CN114549399A

CN114549399A - Liquid level data acquisition method and interface and computer readable storage medium

Info

Publication number: CN114549399A
Application number: CN202111667098.2A
Authority: CN
Inventors: 刘飞虎
Original assignee: Alibaba Cloud Computing Ltd
Current assignee: Alibaba Cloud Computing Ltd
Priority date: 2021-12-31
Filing date: 2021-12-31
Publication date: 2022-05-27

Abstract

The application discloses a liquid level data acquisition method, an interface and a computer readable storage medium. The method comprises the following steps: acquiring a detection image containing a liquid visible area; performing liquid level detection processing on the detection image, and when a liquid level object is detected in the detection image, performing key point detection processing on the detection image to obtain key point information of at least four key points, wherein the key point information comprises pixel coordinates of the key points and key point types corresponding to the pixel coordinates, and the key point types comprise: a starting point and an end point of the liquid visible region, and a liquid level left side point and a liquid level right side point in the liquid visible region; and calculating the liquid level data according to the key point information. The embodiment of the application reduces the complexity of detection, improves the applicability of the liquid level gauge device and simultaneously improves the accuracy of reading.

Description

Liquid level data acquisition method and interface and computer readable storage medium

Technical Field

The present application relates to the field of image processing technologies, and in particular, to a liquid level data obtaining method and interface, and a computer-readable storage medium.

Background

With the development of computer technology, intelligent technology is also increasingly applied to the life and work of people. In particular, in a scenario where a large-scale plant or the like is provided with a large number of instrumentation devices to monitor the operating state of the plant, it is often necessary to manually read each of the instrumentation devices. For example, specialized measurement personnel are required to walk to the location of each meter device and record the meter's data by observing the readings.

For the liquid level type instrument device, since the reading is presented through the liquid level, the industry generally relies on manual reading at present, or reading after the device is converted into an electrical signal. In particular, in the implementation of services such as power patrol, a large number of different types of level gauge reading requirements are encountered, which require direct output of the reading results. However, it is very difficult to obtain accurate reading results with high robustness for various complex real environments.

Disclosure of Invention

The embodiment of the application provides a liquid level data acquisition method, an interface and a computer readable storage medium, so as to overcome the defect of low robustness of liquid level data acquisition in a complex environment in the prior art.

In order to achieve the above object, an embodiment of the present application provides a liquid level data obtaining method, including:

acquiring a detection image containing a liquid visible area;

performing liquid level detection processing on the detection image, and when a liquid level object is detected in the detection image, performing key point detection processing on the detection image to obtain key point information of at least four key points, wherein the key point information comprises pixel coordinates of the key points and key point types corresponding to the pixel coordinates, and the key point types comprise: a starting point and an end point of the liquid visible region, and a liquid level left side point and a liquid level right side point in the liquid visible region, wherein the starting point is a point in the liquid visible region where a coordinate value in a longitudinal axis direction is minimum, and the end point is a point in the liquid visible region where a coordinate value in the longitudinal axis direction is maximum;

and calculating the liquid level data according to the key point information.

The embodiment of the present application further provides a liquid level data acquisition interface, which is used for acquiring liquid level data of a liquid level type meter, wherein the interface includes:

the parameter input area is used for displaying the input control so as to receive parameters which are input by a user and used for acquiring a detection image containing a liquid visible area;

the data display area is used for displaying the liquid level data;

the image display area is used for displaying the detection image and receiving a feedback instruction of a user for the detection image, wherein the image display area further displays at least four key points on the detection image, and the types of the key points comprise: the liquid level display device comprises a starting point and an end point of the liquid visible region, and a liquid level left side point and a liquid level right side point in the liquid visible region, wherein the starting point is a point in the liquid visible region where a coordinate value in the direction of a longitudinal axis is minimum, and the end point is a point in the liquid visible region where the coordinate value in the direction of the longitudinal axis is maximum.

An embodiment of the present application further provides an electronic device, including:

a memory for storing a program;

and the processor is used for operating the program stored in the memory, and the program executes the liquid level data acquisition method provided by the embodiment of the application when running.

Embodiments of the present application further provide a computer-readable storage medium, on which a computer program executable by a processor is stored, where the program, when executed by the processor, implements the liquid level data acquiring method provided in the embodiments of the present application.

According to the liquid level data acquisition method, the interface and the computer readable storage medium, whether the liquid level is included is determined by acquiring the detection image including the liquid visible region and detecting the liquid level of the detection image, and when the liquid level is determined to exist, the key points in the image are detected to obtain at least four pieces of key point information including the starting point, the terminal point and the left and right side points of the liquid level of the liquid visible region, so that the liquid level data can be calculated according to the key point information.

The foregoing description is only an overview of the technical solutions of the present application, and the present application can be implemented according to the content of the description in order to make the technical means of the present application more clearly understood, and the following detailed description of the present application is given in order to make the above and other objects, features, and advantages of the present application more clearly understandable.

Drawings

Various other advantages and benefits will become apparent to those of ordinary skill in the art upon reading the following detailed description of the preferred embodiments. The drawings are only for purposes of illustrating the preferred embodiments and are not to be construed as limiting the application. Also, like reference numerals are used to refer to like parts throughout the drawings. In the drawings:

fig. 1a to fig. 1c are schematic views of application scenarios of a liquid level data acquisition scheme provided in an embodiment of the present application;

FIG. 2 is a flow chart of one embodiment of a liquid level data acquisition method provided herein;

FIG. 3a is a flow chart of another embodiment of a liquid level data acquisition method provided herein;

FIG. 3b is a schematic diagram of an invalid detected image according to an embodiment of the present application;

FIG. 4a is a schematic structural diagram of an embodiment of a liquid level data acquisition device provided by the present application;

FIG. 4b is a schematic view of a liquid level data acquisition interface provided herein;

fig. 5 is a schematic structural diagram of an embodiment of an electronic device provided in the present application.

Detailed Description

Exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be embodied in various forms and should not be limited by the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.

Example one

The scheme provided by the embodiment of the application can be applied to any system with data processing, such as a server system comprising a chip with a data processing function and related components, and the like. Fig. 1a to 1c are schematic views of application scenarios of the liquid level data acquisition scheme provided in the embodiment of the present application, and the scenarios shown in fig. 1a to 1c are merely examples to which the technical solution of the present application is applicable.

In the existing meter reading technology, for a liquid level type meter device, the reading is presented by the liquid level. In the case of level meters, such as those shown in fig. 1a to 1c, the meter face usually has a window therein, and the windows of different types of meters are also different in shape, such as circular, square, irregular, and so on. The level of the liquid is displayed in a window of the level type meter so that a worker or a measurer can observe the level through the window and determine a reading by corresponding to a scale on one side or both sides of the level.

Currently, the industry usually relies on manual reading or reading after converting the signals into electrical signals by using a device. However, in the implementation of services such as power patrol, a large number of different types of reading requirements of the liquid level type meter are met, and due to the complex types, the reading results cannot be directly output through images or video streams independently. Therefore, a reading acquisition scheme of the liquid level type meter, which is applicable to multiple scenes, accurate in result and short in iteration period, is needed.

Therefore, the present embodiment proposes a liquid level data acquisition scheme, as shown in fig. 1a to 1c, which can acquire a detection image containing a liquid visible region from a captured video or image. The test image may be an image containing the complete dial. For example, the dial shown in fig. 1a is circular and has a circular window in the circular outer frame; the dial shown in fig. 1b is rectangular with an oblong window therein; the dial shown generally in fig. 1c is rectangular with a rectangular window therein. The windows are liquid visible areas, liquid levels can be displayed, and liquid level data of the detection image are determined through the corresponding relation between the liquid levels formed in the liquid visible areas and scales on one side of the windows.

According to the liquid level data acquisition scheme provided by the embodiment of the application, the detection image containing the liquid visible region is extracted from the video or image acquired by the liquid level type instrument. For example, in the embodiment of the present application, image frames may be extracted for a video stream acquired by a video acquisition device, such as a camera, for a specified instrument based on an image decoding module or the like. In particular, in the present embodiment, the user may specify the time interval at which the image frames are extracted. For example, the user may specify that one image frame be taken every 1 minute, or may specify that one image frame be taken every 10 seconds. The time interval may be specified as a function of the type of fluid level type instrument being collected or the data. For example, shorter time intervals may be used to extract image frames for a liquid level type instrument with a faster change in liquid level, and longer time intervals may be used to extract image frames for a liquid level type instrument with a slower change in liquid level, thereby ensuring that appropriate image frames are extracted for subsequent liquid level data calculations. After the image frames are extracted at time intervals, the extracted image frames may be subjected to preprocessing, for example, image quality filtering processing may be performed. In embodiments of the present application, the extracted image frames may be processed by employing a filtering module, such as a deep learning classification model. For example, it may be determined whether there is a problem of blurring or blurring in the image frames by determining image features extracted from the image frames in the filtering module, and in particular, it may be determined whether there is a problem of sufficient clarity in the liquid visible region or virtual focus in the liquid visible region, and the image frames with these problems may be discarded without performing subsequent calculation processing, so that subsequent invalid results may be avoided.

After filtering out image frames with low image quality, the image frames may be subjected to tabulated type detection and then may be further filtered. For example, various deep learning models may be used to detect the type corresponding to the dial shown in the image frames, and the image frames that are not suitable for reading calculation may be further filtered out according to whether the size of the detected dial meets the size threshold requirement. For example, images can be filtered that are too small in dial size to actually make the liquid level invisible. Therefore, the type corresponding to the meter and the coordinates of the detection frame identifying the visible region of the liquid can be finally obtained through the processing. For example, in the case of a liquid-visible region of a circular window as shown in fig. 1a, the detection frame may be a rectangular frame tangent to the outline of the circular window; in the case of the liquid-visible region of an oblong window as shown in fig. 1b, the detection frame may be a rectangular frame surrounding the circular window; in the case of a liquid-visible region of a rectangular window as shown in fig. 1c, the detection frame may be a rectangular frame surrounding the rectangular window. Of course, in the present embodiment, other shapes may be adopted as long as the shape can surround the liquid visible region.

After the coordinates of the detection frame are obtained, the liquid level in the detection frame can be detected. For example, it may be detected whether or not the liquid surface object is present in the detection frame. In this process, it is possible to determine whether or not a liquid level object is present in the detection frame for the image features extracted from the image frame by, for example, deep learning a classification model.

When a liquid level object is detected in the detection frame, that is, when a liquid level is visible in the image frame, a keypoint detection process may be performed on the detection image to obtain keypoint information of at least four keypoints. In the embodiment of the present application, the key points may be a start point, an end point of the liquid visible region, a liquid level left side point and a liquid level right side point in the liquid visible region. For example, the starting point of the liquid visible region is the point with the smallest ordinate in the liquid visible region, i.e., the lowest point of the longitudinal axis of the liquid visible region. For example, in the scenario shown in FIG. 1a, the starting point may be the point where the lower contour of the circular liquid-visible region is tangent to the detection box; in the scenario shown in fig. 1b, the starting point may be a point where the lower contour of the liquid visible region is tangent to the detection box; in the scenario shown in fig. 1c, the starting point may be any point of the underlying contour of the liquid-visible area. Similarly, the end point of the liquid-visible region is the point of the liquid-visible region where the ordinate is the largest, i.e., the uppermost point of the liquid-visible region on the ordinate. For example, in the scenario shown in fig. 1a, the end point may be the point where the upper contour of the circular liquid visible area is tangent to the detection box; in the scenario shown in fig. 1b, the end point may be a point where the upper contour of the liquid visible area is tangent to the detection box; in the scenario shown in fig. 1c, the end point may be any point of the upper contour of the visible area of the liquid. Furthermore, the leftmost and rightmost points of the liquid level in the visible region of the liquid can also be determined by feature extraction. For example, in a scenario as shown in fig. 1a to 1c, the left point may be a point where the liquid level intersects the left contour of the liquid-visible region, and the right point may be a point where the liquid level intersects the right contour of the liquid-visible region.

Therefore, the coordinates of the four key points in the liquid-visible region can be further calculated from the above-detected four key points. After obtaining the coordinates, the distribution relationship of the four key points in the visible liquid area can be calculated first, so as to verify whether the detection of the four key points is reasonably effective. For example, the y-axis coordinate of the liquid level, i.e., the coordinate of the liquid level in the longitudinal axis direction in the liquid-visible region, may be calculated from the coordinates of the left-side point and the right-side point of the liquid level among the calculated key points, and compared with the coordinates of the end point of the liquid-visible region, i.e., the point of the uppermost contour of the liquid-visible region. When the comparison result shows that the coordinate of the liquid level on the vertical axis is higher than the coordinate of the end point of the liquid visible region, the detection error of the point on the left side or the right side of the liquid level or the detection error of the end point of the liquid visible region is indicated. That is, the output result of the above-described detection processing of the key point is not reasonable. Similarly, if it is determined from the detected coordinates of the key point that the difference between the y-axis coordinates of the left point of the liquid level and the right point of the liquid level is too large, for example, greater than a certain threshold, it may be considered that the determination of the key point is not reasonable. The coordinates of these four key points cannot be used for subsequent level data calculations.

Conversely, if it is determined from the above comparison that there is no unreasonable situation in the coordinates of the key points, the coordinates of the midpoint of the liquid level can be calculated from the coordinates of the points on the left and right sides of the liquid level, and the difference in the y-axis coordinates of the midpoint from the starting point or midpoint of the liquid-visible region can be used as the reading value of the liquid level, i.e., the liquid level data.

Further, when the liquid level is not detected in the above liquid level object detection processing, that is, the liquid level is not visible in the image frame, that is, the liquid level is not visible in the liquid visible region, the case may be that the liquid in the dial is in a full scale state or an empty scale state at the time of capturing the detection image. I.e. the liquid either fills the meter or the liquid is already below the lowest contour plane of the liquid visible area of the meter. Therefore, in this case, it is possible to further perform liquid object detection to confirm whether the liquid level is invisible, or whether the liquid in the dial is in a full scale state or an empty scale state. For example, it is possible to determine whether or not a liquid object exists in the liquid-visible region by extracting an image feature for the detection image or for the liquid-visible region and based on the extracted feature. When the liquid object is judged not to exist, the condition that the liquid level is invisible can be actually judged, namely the state when the detection image is collected is that the liquid in the instrument is in an empty scale state. On the contrary, when the liquid object is judged to exist, the condition that the liquid level is invisible can be judged, namely the state when the detection image is collected is that the liquid in the dial plate is in a full scale state.

Therefore, the liquid level object can be determined from the detection image without the embodiment of the application, the key points are extracted from the detection image, and the liquid level data can be calculated according to the coordinates of the key points. Since the liquid visible region is usually a transparent window and the surrounding dial region is a metal or other opaque entity, the calculation for determining the edge of the liquid visible region is not only relatively simple, but also highly accurate, compared to the calculation for identifying the same transparent liquid level in the liquid visible region according to the image characteristics. And the adaptability to the types of dials and liquid-visible areas is also high.

According to the liquid level data acquisition scheme provided by the embodiment of the application, whether the liquid level is contained is determined by acquiring the detection image containing the liquid visible region and detecting the liquid level of the detection image, the key points in the image are detected when the liquid level is determined to exist, at least four pieces of key point information including the starting point, the end point and the left and right side points of the liquid level in the liquid visible region are obtained, then the liquid level data can be calculated according to the key point information, compared with the prior art that the liquid level needs to be identified by using image features, the complexity of detection is reduced, the applicability to liquid level type meter equipment is improved, and meanwhile, the accuracy of reading is improved.

The foregoing embodiments are illustrative of technical principles and exemplary application frameworks of the embodiments of the present application, and specific technical solutions of the embodiments of the present application are further described in detail below through a plurality of embodiments.

Example two

Fig. 2 is a flowchart of an embodiment of the liquid level data obtaining method provided in the present application, and an execution subject of the method may be various terminal or server devices with data processing capability, or may be a device or chip integrated on these devices. As shown in fig. 2, the liquid level data acquisition method includes the following steps:

s201, acquiring a detection image containing a liquid visible area.

The detection image including the liquid visible region may be acquired in step S201 as an object of subsequent processing. For example, a detection image containing a visible region of liquid, such as shown in fig. 1a to 1c, may be acquired by extracting an image frame from a video or image taken for a liquid type apparatus. The detection image may be an image including a complete dial, and there may be an area in the dial where the liquid can be seen from the outside, i.e., a liquid visible area, and the liquid level data of the detection image is determined by the correspondence of the liquid level in the liquid visible area and the scale on the window side.

S202, liquid level detection processing is carried out on the detection image, and when a liquid level object is detected in the detection image, key point detection processing is carried out on the detection image to obtain key point information of at least four key points.

The liquid level in the detection image may be detected at step S202. And detecting whether the liquid level object exists in the detection image or not. In this process, it is possible to determine whether or not a liquid level object is present in the detection image with respect to the image features extracted from the image frame by, for example, deep learning a classification model.

When a liquid level object is detected in the detection image, that is, when the liquid level is visible in the detection image, a key point detection process may be further performed on the detection image in step S202 to obtain key point information of at least four key points.

In this embodiment of the present application, the keypoint information may include pixel coordinates of the keypoint and a keypoint type corresponding to the pixel coordinates, and the keypoint type includes: the starting point and the end point of the liquid visible region, and the left-side point and the right-side point of the liquid surface in the liquid visible region. That is, in the present embodiment, the key points detected in step S202 may be the start point, the end point, the liquid level left side point, and the liquid level right side point in the liquid visible region. In other words, it is practical to detect four points on the boundary of the liquid-visible region of the liquid-level type apparatus, i.e., the highest point and the lowest point and two intersections of the liquid level therein on the boundary of the liquid-visible region, i.e., the left-side point and the right-side point, in step S202. Therefore, in the embodiment of the application, the key point can be determined by detecting the liquid visible area in the detection image and the edge of the inner frame of the dial plate.

In the embodiment of the present application, the starting point of the liquid-visible region may be a point lowermost of the longitudinal axis of the liquid-visible region. For example, the starting point may be a point where the lower contour of the circular liquid-visible area of the dial shown in fig. 1a is tangent to a detection frame surrounding the liquid-visible area. Similarly, the end point of the liquid-visible region may be the uppermost point of the longitudinal axis of the liquid-visible region. For example, in the scenario shown in fig. 1a, the end point may be the point where the upper contour of the circular liquid visible area is tangent to the detection box. Furthermore, the leftmost and rightmost points of the liquid level in the visible region of the liquid can also be determined by feature extraction. For example, in a scenario as shown in fig. 1a, the left point may be a point where the liquid level intersects the left contour of the liquid-visible region, and the right point may be a point where the liquid level intersects the right contour of the liquid-visible region.

And S203, calculating liquid level data according to the key point information.

In step S203, level data may be calculated from the keypoints detected in step S202. In the embodiment of the present application, since the boundary of the liquid visible region has been determined by the calculated key points in step S202, and the intersection of the liquid level on the left and right boundaries is also determined, it is possible to determine the total height of the liquid visible region from the top point and the bottom point of the liquid visible region in step S203, and determine the position of the liquid level from the boundary points on the left and right sides of the liquid level, whereas in general, the total height of the liquid visible region, that is, the total length of the graduated scale corresponding to the liquid level, and thus the liquid level data is determined by calculating the ratio of the distance of the liquid level from the top point or the bottom point to the total height of the liquid visible region.

According to the liquid level data acquisition method provided by the embodiment of the application, whether the liquid level is contained is determined by acquiring the detection image containing the liquid visible region and detecting the liquid level of the detection image, the key points in the image are detected when the liquid level is determined to exist, so that at least four pieces of key point information including the starting point, the end point and the left and right side points of the liquid visible region are obtained, and then the liquid level data can be calculated according to the key point information.

EXAMPLE III

Fig. 3a is a flow chart of another embodiment of the data acquisition method of the liquid level type meter provided in the present application, and the execution subject of the method may be various terminal or server devices with data processing capability, or may be devices or chips integrated on these devices. As shown in fig. 3a, on the basis of the embodiment shown in fig. 2, the data acquisition method of the liquid level type meter provided by the embodiment of the present application may include the following steps:

s301, a detection image containing a liquid visible area is acquired.

In step S301, a detection image including a liquid visible region is acquired. For example, in this step, at least one frame of original image may be obtained first, and the image quality detection process may be performed on the original image by using the third deep learning classification model. In particular, a detection image containing a visible region of the liquid may be extracted from a video or image captured for a liquid level type instrument. For example, in the instrument scenario shown in fig. 1a, the fluid level instrument shown therein may have a circular dial, and the dial may include a circular window, through which the internal fluid may be displayed, so that the window may be a fluid visible region of the dial, and a graduated scale may be provided on the dial on one side of the window, so that fluid level data may be determined according to the correspondence between the fluid level and the scale.

Therefore, in the embodiment of the present application, image frames may be extracted for a video stream acquired by a video acquisition device such as a camera for a specified instrument based on, for example, an image decoding module in step S301. In particular, in the present embodiment, the user may specify the time interval at which the image frames are extracted in step S301. For example, the user may specify that one image frame be taken every 1 minute, or may specify that one image frame be taken every 10 seconds.

The time interval used in step S301 may be a time interval set according to the type of the collected liquid level type instrument or a data change. For example, a shorter time interval may be used to extract image frames for a liquid level type instrument with a faster liquid level change, and a longer time interval may be used to extract image frames for a liquid level type instrument with a slower liquid level change, so that it may be ensured that a suitable image frame is extracted in step S301 for the subsequent liquid level data calculation.

Further, when the image frames are extracted at time intervals in step S301, the extracted image frames may be pre-processed first. For example, image quality filtering processing may be performed. In embodiments of the present application, the extracted image frames may be processed by employing a filtering module, such as a deep learning classification model. For example, it may be determined whether there is a problem of blurring or blurring in the image frames extracted in step S301, especially whether there is a problem of sufficient clarity in the liquid visible region or whether there is virtual focus in the liquid visible region, by determining image features extracted from the image frames in the filtering module, and the image frames with these problems may be discarded without performing subsequent calculation processing, so that subsequent invalid results may be avoided.

After filtering out the image frames with low image quality, in step S301, the image frames may also be subjected to meter type detection, and then the image frames may also be further filtered. For example, various deep learning models may be used to detect the type corresponding to the dial shown in the image frame extracted in step S301, and image frames that are not suitable for reading calculation may be further filtered out according to whether the size of the detected dial meets the size threshold requirement.

For example, images can be filtered that are too small in dial size to actually make the liquid level invisible. Therefore, the type corresponding to the meter and the coordinates of the detection frame identifying the visible region of the liquid can be finally obtained through the processing. For example, as shown in fig. 1a, in the case of a liquid-visible region of a circular window as shown in fig. 1a, the detection frame may be a rectangular frame tangent to the outline of the circular window. Of course, in the present embodiment, other shapes may be adopted as long as the shape can surround the liquid visible region.

In addition, in step S301, the extracted image frame may be further subjected to shooting orientation detection. For example, the meter dial features in the image frames may be extracted by various depth learning models, and it is determined whether a shooting subject corresponding to the extracted image frame, such as a camera or a camera, is tilted or rotated during shooting, so that the meters in the image frame are not at the front-view shooting angle, and therefore, in this case, the image frame may be adjusted according to the rotation angle of the meters in the extracted image frame with respect to a reference meter image stored in advance, for example, the image frame or the meter area in the image frame may be rotated as a whole so that the meters in the rotated image frame are at the front-view position.

And S302, performing liquid level detection processing on the detection image by adopting a second deep learning classification model.

After the coordinates of the detection frame are obtained in step S301, the liquid level in the detection frame determined in step S301 may be detected in step S302. For example, it may be confirmed whether or not a liquid level object exists in the detection frame determined in step S301. In step S302, it may be determined whether or not a liquid level object exists in the detection frame for the image features extracted from the image frame by a deep learning classification model.

Alternatively, the detection image may be subjected to a binarization process to obtain a grayscale image, the grayscale image may be subjected to a binarization process, and the presence or absence of the liquid surface object in the detection image may be determined based on the result of the binarization process. Since the liquid is usually observed, the visible region of the liquid is usually made of transparent material, and the liquid therein is usually air above, and the refractive index of the air is different from that of the liquid. But because of the transparency property, it is difficult in the prior art to distinguish liquid from air in the visible area of the liquid by extracting image features. In the embodiment of the present application, in step S302, the detected image may be grayed, for example, converted into an object with different grayscale values, in which case, the material of the visible region of the liquid and the liquid therein have different grayscale values respectively due to different refractive indexes to the light, so that it can be determined whether there is a boundary between the liquid and the air, that is, the liquid level, by performing binarization on the grayscale image through the numerical change after binarization.

And S303, when the liquid level object is detected in the detection image, carrying out key point detection processing on the detection image to obtain key point information of at least four key points.

When the presence of the liquid level is detected, that is, the liquid level is visible in the detection image, the key point detection processing may be further performed on the detection image in step S303 to obtain key point information of at least four key points. Specifically, a pre-trained key point detection model may be adopted to perform key point detection processing on the detected image, so as to obtain key point information of at least four key points.

In this embodiment of the present application, the keypoint information may include pixel coordinates of the keypoint and a keypoint type corresponding to the pixel coordinates, and the keypoint type includes: the starting point and the end point of the liquid visible region, and the left-side point and the right-side point of the liquid surface in the liquid visible region. That is, in the present embodiment, the key points detected in step S303 may be the start point and the end point of the liquid-visible region, the liquid-level left point and the liquid-level right point in the liquid-visible region. In other words, four points on the boundary of the liquid-visible region of the liquid level type apparatus, i.e., the highest point and the lowest point and two intersections of the liquid level therein on the boundary of the liquid-visible region, i.e., the left point and the right point, are actually detected in step S303. Therefore, in the embodiment of the application, the key point can be determined by detecting the liquid visible area in the detection image and the edge of the inner frame of the dial plate.

In the embodiment of the present application, the starting point of the liquid-visible region may be a lowermost point of a longitudinal axis of the liquid-visible region. For example, the starting point may be a point where the lower contour of the circular liquid-visible area of the dial shown in fig. 1a is tangent to a detection frame surrounding the liquid-visible area. Similarly, the end point of the liquid-visible region may be the uppermost point of the longitudinal axis of the liquid-visible region. For example, in the scenario shown in fig. 1a, the end point may be the point where the upper contour of the circular liquid visible area is tangent to the detection box. Furthermore, the leftmost and rightmost points of the liquid level in the visible region of the liquid can also be determined by feature extraction. For example, in a scenario as shown in fig. 1a, the left point may be a point where the liquid level intersects the left contour of the liquid-visible region, and the right point may be a point where the liquid level intersects the right contour of the liquid-visible region.

And S304, when the liquid level object is not detected in the detection image, performing liquid detection processing on the detection image by adopting a first deep learning classification model.

S305, when the liquid object is not detected in the detection image, it is determined that the liquid level data is zero.

S306, when the liquid object is detected in the detection image, the liquid level data is determined to be a preset value.

Further, when the liquid level object cannot be detected in step S302, that is, it is determined that the liquid level is not visible in the image frame, that is, the liquid level is not visible in the liquid visible region, then the case may be that the liquid in the dial is in the full scale state or the empty scale state at the time of capturing the detection image. I.e. the liquid either fills the meter or the liquid is already below the lowest contour plane of the liquid visible area of the meter.

Therefore, in step S304, it is possible to further perform liquid object detection to confirm whether the liquid level is invisible, or whether the liquid in the dial is in a full scale state or an empty scale state. For example, it is possible to determine whether or not a liquid object is present in the liquid-visible region in step S305 by extracting image features for the detection image or for the liquid-visible region and from the extracted features. In step S305, when it is determined that there is no liquid object, it can be actually determined that the liquid level is not visible, that is, the state when the detection image is acquired is that the liquid in the instrument is in an empty scale state, and therefore, the liquid level data can be directly set to zero in step S305. On the contrary, in step S306, when it is determined that the liquid object exists in the detection image, it may be determined that the liquid level determined in step S304 is not visible, that is, the state when the detection image is captured is the state where the liquid in the dial is in the full scale state. Thus, the level data may be directly set to the maximum value of the graduated scale, for example, in step S306, which may indicate that the reading of the level-type instrument is maximum.

And S307, acquiring the pixel coordinate of the middle point of the liquid level according to the pixel coordinates of the left point and the right point of the liquid level.

And S308, calculating liquid level data according to the proportion of the projection distance of the middle point of the liquid level to the projection distance of the starting point and the end point in the longitudinal axis direction of the visible area of the liquid level.

In steps S307 and S308, liquid level data may be calculated from the key points detected in step S303. Specifically, in the embodiment of the present application, since the boundary of the liquid visible region has been determined by the calculated key point in step S303, and the intersection point of the liquid level on the left and right boundaries has also been determined, it is possible to determine the total height of the liquid visible region from the top point and the bottom point of the liquid visible region and determine the position of the liquid level by the boundary points on the left and right sides of the liquid level in step S307, whereas in general, the total height of the liquid visible region, that is, the total length of the graduated scale corresponding to the liquid level, and thus the liquid level data is determined by calculating the ratio of the distance of the liquid level from the top point or the bottom point to the total height of the liquid visible region.

For example, based on the coordinates of the key points obtained in step S303, the distribution relationship of the four key points in the liquid visible region may be first calculated, so as to verify whether the detection of the four key points is reasonably effective. For example, the pixel coordinates of the point on the left side of the liquid surface or the point on the right side of the liquid surface may be compared with the pixel coordinates of the end point of the liquid-visible region. When the coordinate value of the left point or the right point of the liquid surface in the direction of the vertical axis is higher than the coordinate value of the end point in the direction of the vertical axis, that is, when the comparison result indicates that the coordinate of the liquid surface on the vertical axis is higher than the coordinate of the end point of the visible region of the liquid, it indicates that the detection of the left point or the right point of the liquid surface is wrong, or the detection of the end point of the visible region of the liquid is wrong. That is, the output result of the detection processing of the above-described key point is not reasonable, and therefore, it is determined that the detection image is invalid. Similarly, if it is determined that the projection distance between the left point of the liquid surface and the right point of the liquid surface in the longitudinal axis direction is greater than the preset threshold according to the detected coordinates of the key points, that is, the difference between the y-axis coordinates of the left point of the liquid surface and the right point of the liquid surface is too large, it can also be said that the determination of the key points is not reasonable, and therefore the coordinates of the four key points cannot be used for subsequent calculation of the liquid level data. Fig. 3b is a schematic diagram of an invalid detection image according to an embodiment of the present application. As shown in fig. 3b, in this example, the y-axis coordinate of the point on the left side of the page is significantly higher than the y-axis coordinate of the point on the right side of the page, and therefore, the detected image is determined to be invalid. For the detected image determined to be invalid, the detected image and corresponding key point information can be used as training data and input into the key point detection model so as to perform iterative training on the key point detection model.

Conversely, if it is determined that there is no unreasonable situation in the coordinates of the key points based on the above comparison, the reading value of the liquid level, i.e., the liquid level data, can be calculated based on the difference in the y-axis coordinates of the midpoint coordinates of the liquid level and the start and end points of the liquid visible region calculated in step S307.

Example four

Fig. 4a is a schematic structural diagram of an embodiment of a liquid level data acquiring apparatus provided in the present application, which may be used for executing the liquid level data acquiring method shown in fig. 2 or fig. 3. As shown in fig. 4a, the liquid level data acquiring device may include: an acquisition module 41, a detection module 42 and a calculation module 43.

The acquisition module 41 may be used to acquire a detection image containing a visible region of the liquid.

The acquisition module 41 may acquire the detection image including the liquid visible region as an object of the subsequent processing. For example, a detection image containing a visible region of liquid, such as shown in fig. 1, may be acquired by extracting an image frame from a video or image taken for a liquid type apparatus. The detection image may be an image including a complete dial, and there may be an area in the dial where the liquid can be seen from the outside, i.e., a liquid visible area, and the liquid level data of the detection image is determined by the correspondence of the liquid level in the liquid visible area and the scale on the window side.

The obtaining module 41 may first obtain at least one frame of original image and perform image quality detection processing on the original image by using the third deep learning classification model. In particular, a detection image containing a visible region of the liquid may be extracted from a video or image captured for a level-type instrument. For example, in the instrument scenario shown in fig. 1, the fluid level type instrument shown therein may have a circular dial plate, and the dial plate includes a circular window through which the internal fluid can be displayed, so that the window can be a fluid visible region of the dial plate, and a graduated scale may be provided on the dial plate on one side of the window, so that fluid level data can be determined according to the correspondence of fluid level and scale.

Therefore, in the embodiment of the present application, the acquisition module 41 may extract image frames for a video stream acquired by a video acquisition device, such as a camera, for a specified instrument based on, for example, an image decoding module. In particular, in the embodiment of the present application, the user may specify the time interval at which the image frame is extracted by the acquisition module 41. For example, the user may specify that one image frame be taken every 1 minute, or may specify that one image frame be taken every 10 seconds.

The time interval used by the acquisition module 41 may be a time interval set according to the type of fluid level type instrument being collected or the data change. For example, a shorter time interval may be used to extract image frames for a liquid level type instrument with a faster liquid level change, and a longer time interval may be used to extract image frames for a liquid level type instrument with a slower liquid level change, so that it may be ensured that the acquisition module 41 extracts a suitable image frame for the subsequent liquid level data calculation.

In addition, when the obtaining module 41 extracts the image frames at time intervals, the extracted image frames may be preprocessed first. For example, image quality filtering processing may be performed. In the embodiment of the present application, the extracted image frames may be processed by employing, for example, a deep learning classification model. For example, it may be determined whether there is a problem of blurring or blurring in the image frames extracted by the acquisition module 41 by determining image features extracted from the image frames, and in particular, it may be determined whether there is a problem of sufficient clarity in the liquid visible region or virtual focus in the liquid visible region, and the image frames with these problems may be discarded without performing subsequent calculation processing, so that subsequent invalid results may be avoided.

After filtering out the image frames with low image quality, the obtaining module 41 may also perform tabulated type detection on the image frames, and then may further screen the image frames. For example, various deep learning models may be used to detect the type corresponding to the dial shown in the image frames extracted by the acquisition module 41, and the image frames that are not suitable for reading calculation may be further filtered out according to whether the detected size of the dial meets the size threshold requirement.

For example, images can be filtered that are too small in dial size to actually make the liquid level invisible. Therefore, the type corresponding to the meter and the coordinates of the detection frame identifying the visible region of the liquid can be finally obtained through the processing. For example, as shown in fig. 1, in the case of a liquid-visible region of a circular window as shown in fig. 1, the detection frame may be a rectangular frame that is tangent to the outline of the circular window. Of course, in the present embodiment, other shapes may be adopted as long as the shape can surround the liquid visible region.

The detection module 42 may be configured to perform liquid level detection processing on the detection image, and when a liquid level object is detected in the detection image, perform key point detection processing on the detection image to obtain key point information of at least four key points.

The detection module 42 may detect a liquid level in the detection image. And detecting whether the liquid level object exists in the detection image or not. In this process, it is possible to determine whether or not a liquid level object is present in the detection image with respect to the image features extracted from the image frame by, for example, deep learning a classification model.

When a liquid level object is detected in the detection image, that is, when the liquid level is visible in the detection image, the detection module 42 may further perform a key point detection process on the detection image to obtain key point information of at least four key points.

In this embodiment of the present application, the keypoint information may include pixel coordinates of the keypoint and a keypoint type corresponding to the pixel coordinates, and the keypoint type includes: the starting point and the end point of the liquid visible region, and the left-side point and the right-side point of the liquid surface in the liquid visible region. That is, in the embodiment of the present application, the key points detected by the detection module 42 may be a start point and an end point of the liquid visible region, a left-side point and a right-side point of the liquid surface in the liquid visible region. In other words, the detection module 42 actually detects four points on the boundary of the liquid-visible region of the liquid-level type apparatus, i.e., the highest point and the lowest point and two intersections of the liquid level therein on the boundary of the liquid-visible region, i.e., the left-side point and the right-side point. Therefore, in the embodiment of the application, the key point can be determined by detecting the liquid visible area in the detection image and the edge of the inner frame of the dial plate.

In the embodiment of the present application, the starting point of the liquid-visible region may be a point lowermost of the longitudinal axis of the liquid-visible region. For example, the starting point may be a point where the lower contour of the circular liquid-visible area of the dial shown in fig. 1 is tangent to a detection frame surrounding the liquid-visible area. Similarly, the end point of the liquid-visible region may be the uppermost point of the longitudinal axis of the liquid-visible region. For example, in the scenario shown in fig. 1, the end point may be a point where the upper contour of the circular liquid-visible region is tangent to the detection box. Furthermore, the leftmost and rightmost points of the liquid level in the visible region of the liquid can also be determined by feature extraction. For example, in a scenario as shown in fig. 1, the left point may be a point where the liquid level intersects the left contour of the liquid-visible region, and the right point may be a point where the liquid level intersects the right contour of the liquid-visible region.

Specifically, the detection module 42 may perform a graying process on the detection image to obtain a grayscale image, perform a binarization process on the grayscale image, and determine whether the liquid level object exists in the detection image according to a binarization processing result. Since the liquid is usually observed, the visible region of the liquid is usually made of transparent material, and the liquid therein is usually air above, and the refractive index of the air is different from that of the liquid. But because of the transparency property, it is difficult in the prior art to distinguish liquid from air in the visible area of the liquid by extracting image features. In the embodiment of the present application, the detection module 42 may perform graying on the detection image, for example, convert the detection image into an object with different grayscale values, in which case, the material of the visible region of the liquid and the liquid therein have different grayscale values respectively due to different refractive indexes to the light, so that whether a boundary, i.e., a liquid level, exists between the liquid and the air can be determined by binarizing the grayscale image through a change of the binarized value.

When the detection module 42 detects that there is a liquid level by graying the image and using the deep learning model, that is, when the liquid level is visible in the detection image, the detection module may further perform a key point detection process on the detection image to obtain key point information of at least four key points.

Further, when the detection module 42 fails to detect a level object, that is, determines that the level is not visible in the image frame, that is, the level is not visible in the liquid visible region, then the situation may be that the liquid in the dial plate is in a full scale state or an empty scale state at the time of acquiring the detection image. I.e. the liquid either fills the meter or the liquid is already below the lowest contour plane of the liquid visible area of the meter.

Therefore, the detection module 42 can further perform liquid object detection to confirm whether the liquid level is invisible, or whether the liquid in the dial is in a full scale state or an empty scale state. For example, it is possible to determine whether or not a liquid object exists in the liquid-visible region by extracting an image feature for the detection image or for the liquid-visible region and based on the extracted feature. When the liquid object does not exist, the condition that the liquid level is invisible can be actually judged, namely the state when the detection image is collected is that the liquid in the instrument is in an empty scale state, so that the liquid level data can be directly set to be zero. On the contrary, when the liquid object is judged to exist in the detection image, the condition that the determined liquid level is invisible can be judged, namely the state when the detection image is collected is that the liquid in the dial plate is in a full scale state. Thus, the level data can be set directly to the maximum value of, for example, a graduated scale, and thus can indicate that the reading of the level-type instrument is maximum.

The calculation module 43 may be configured to calculate the liquid level data based on the keypoint information.

The calculation module 43 may calculate the liquid level data based on the keypoints detected by the detection module 42. In the embodiment of the present application, since the detection module 42 has determined the boundary of the liquid visible region by the calculated key point and also determines the intersection point of the liquid level on the left and right boundaries, the calculation module 43 can determine the total height of the liquid visible region according to the top point and the bottom point of the liquid visible region and determine the position of the liquid level by the boundary points on the left and right sides of the liquid level, whereas in general, the total height of the liquid visible region is also the total length of the graduated scale corresponding to the liquid level, and therefore, the liquid level data is determined by calculating the ratio of the distance of the liquid level from the top point or the bottom point to the total height of the liquid visible region.

The liquid level data acquisition device that this application embodiment provided, through obtaining the detection image that contains the visible region of liquid, and carry out liquid level detection to this detection image and confirm whether contain the liquid level, detect the key point in this image when confirming that there is the liquid level, with at least four key point information including the visible region's of liquid starting point, terminal point and liquid level left and right side point, and then can calculate liquid level data according to these key point information, compare with the prior art that needs use image feature recognition liquid level, the complexity of detection has been reduced, the suitability to liquid level type table meter equipment has been improved, the accuracy of reading has been promoted simultaneously.

The liquid level data may be calculated based on the keypoints detected by the detection module 42. Specifically, in the embodiment of the present application, since the detection module 42 has determined the boundary of the liquid visible region by the calculated key point and also determines the intersection point of the liquid level on the left and right boundaries, the calculation module 43 can determine the total height of the liquid visible region according to the top point and the bottom point of the liquid visible region and determine the position of the liquid level by the boundary points on the left and right sides of the liquid level, whereas in general, the total height of the liquid visible region is also the total length of the graduated scale corresponding to the liquid level, and therefore, the liquid level data is determined by calculating the ratio of the distance of the liquid level from the top point or the bottom point to the total height of the liquid visible region.

For example, the distribution relationship of the four key points in the visible region of the liquid may be first calculated based on the coordinates of the key points obtained by the detection module 42, so as to verify whether the detection of the four key points is reasonably effective. For example, the calculation module 43 may calculate the pixel coordinates of the midpoint of the liquid level from the coordinates of the left-side point and the right-side point of the liquid level among the calculated key points. And the coordinate in the direction of the longitudinal axis of the liquid level in the liquid-visible region is compared with the coordinate of the end point of the liquid-visible region, i.e. the point of the uppermost contour of the liquid-visible region, on the basis of the y-axis coordinate of this pixel coordinate. When the comparison result shows that the coordinate of the liquid level on the vertical axis is higher than the coordinate of the end point of the liquid visible region, the detection error of the point on the left side or the right side of the liquid level or the detection error of the end point of the liquid visible region is indicated. That is, the output result of the above-described detection processing of the key point is not reasonable. Similarly, if it is determined from the detected coordinates of the key point that the difference between the y-axis coordinates of the left point of the liquid level and the right point of the liquid level is too large, for example, greater than a certain threshold, it may be considered that the determination of the key point is not reasonable. The coordinates of these four key points cannot be used for subsequent level data calculations.

Conversely, if it is determined from the comparison that there is no unreasonable situation in the coordinates of the key points, the difference between the coordinates of the midpoint of the liquid level and the y-axis coordinates of the start or midpoint of the visible region of the liquid calculated by the calculation module 43 can be used as the reading value of the liquid level, i.e., the liquid level data.

FIG. 4b is a schematic view of a fluid level data acquisition interface provided herein. As shown in fig. 4b, the liquid level data acquisition interface may be displayed on a terminal used by the user. For example, the interface may include a parameter input area, a data display area, and an image display area.

The parameter input area may be used to display input controls. Parameters input by a user and used for acquiring a detection image containing a liquid visible area can be received through the input control. For example, the user may set the time interval for image acquisition depending on the type of fluid level type instrument or data changes. The interval time of the extraction of the control frames reaches the detection frequency of the liquid level type meter, for example, the lower frame extraction frequency is corresponding to the scene with slower liquid level change, otherwise, the frame extraction frequency is increased, so that the reading algorithm of the whole meter can be suitable for different application scenes.

The data display area may be used to display the liquid level data.

The image display area can be used for displaying the detection image and receiving a feedback instruction of a user for the detection image.

In this embodiment of the application, the user can view the acquired liquid level data, that is, the reading of the required liquid level type device, through the data display area. The user can also view the acquired original image, the determined detection image and the like in real time through the image display area so as to know the progress of the liquid level data acquisition processing according to the embodiment of the application. In addition, the image display area may further display at least four key points on the detection image, and may further display connecting lines between the key points to show a layout relationship between the key points.

The types of keypoints may include: the starting point and the end point of the liquid visible region, and the left-side point and the right-side point of the liquid surface in the liquid visible region. In particular, in the embodiment of the present application, the image display section may further receive a user's feedback on the displayed image or the key point in a state where the key point on the detection image or the monitor image is displayed. For example, if the displayed original image or detected image is not clear enough from the user's view, or it is determined from the user's experience that there is an error in the original image or detected image used this time, the user can input feedback through the image display area, and thus the image can be discarded. Further, the user may confirm the key points displayed on the detection image, and for example, the position of the key points may be adjusted by touching and dragging the key points.

In addition, the image display area may further display a rectangular detection frame identifying the liquid visible region on the detection image, and the user may also adjust the detection frame by touching and dragging.

EXAMPLE five

The internal functions and structure of the data acquisition device of the gauge of the level type, which can be implemented as an electronic apparatus, have been described above. Fig. 5 is a schematic structural diagram of an embodiment of an electronic device provided in the present application. As shown in fig. 5, the electronic device includes a memory 51 and a processor 52.

The memory 51 stores programs. In addition to the above-described programs, the memory 51 may also be configured to store other various data to support operations on the electronic device. Examples of such data include instructions for any application or method operating on the electronic device, contact data, phonebook data, messages, pictures, videos, and so forth.

The memory 51 may be implemented by any type or combination of volatile or non-volatile memory devices, such as Static Random Access Memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic or optical disks.

The processor 52 is not limited to a Central Processing Unit (CPU), but may be a processing chip such as a Graphic Processing Unit (GPU), a Field Programmable Gate Array (FPGA), an embedded neural Network Processor (NPU), or an Artificial Intelligence (AI) chip. And a processor 52, coupled to the memory 51, for executing the program stored in the memory 51 to execute the liquid level data obtaining method of the second or third embodiment.

Further, as shown in fig. 5, the electronic device may further include: communication components 53, power components 54, audio components 55, display 56, and other components. Only some of the components are schematically shown in fig. 5, and the electronic device is not meant to include only the components shown in fig. 5.

The communication component 53 is configured to facilitate wired or wireless communication between the electronic device and other devices. The electronic device may access a wireless network based on a communication standard, such as WiFi, 3G, 4G, or 5G, or a combination thereof. In an exemplary embodiment, the communication component 53 receives a broadcast signal or broadcast related information from an external broadcast management system via a broadcast channel. In an exemplary embodiment, the communication component 53 further comprises a Near Field Communication (NFC) module to facilitate short-range communication. For example, the NFC module may be implemented based on Radio Frequency Identification (RFID) technology, infrared data association (IrDA) technology, Ultra Wideband (UWB) technology, Bluetooth (BT) technology, and other technologies.

A power supply component 54 provides power to the various components of the electronic device. The power components 54 may include a power management system, one or more power sources, and other components associated with generating, managing, and distributing power for an electronic device.

The audio component 55 is configured to output and/or input audio signals. For example, the audio component 55 includes a Microphone (MIC) configured to receive external audio signals when the electronic device is in an operational mode, such as a call mode, a recording mode, and a voice recognition mode. The received audio signal may further be stored in the memory 51 or transmitted via the communication component 53. In some embodiments, audio assembly 55 also includes a speaker for outputting audio signals.

The display 56 includes a screen, which may include a Liquid Crystal Display (LCD) and a Touch Panel (TP). If the screen includes a touch panel, the screen may be implemented as a touch screen to receive an input signal from a user. The touch panel includes one or more touch sensors to sense touch, slide, and gestures on the touch panel. The touch sensor may not only sense the boundary of a touch or slide action, but also detect the duration and pressure associated with the touch or slide operation.

Those of ordinary skill in the art will understand that: all or a portion of the steps of implementing the above-described method embodiments may be performed by hardware associated with program instructions. The program may be stored in a computer-readable storage medium. When executed, the program performs steps comprising the method embodiments described above; and the aforementioned storage medium includes: various media that can store program codes, such as ROM, RAM, magnetic or optical disks.

Finally, it should be noted that: the above embodiments are only used to illustrate the technical solution of the present invention, and not to limit the same; while the invention has been described in detail and with reference to the foregoing embodiments, it will be understood by those skilled in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some or all of the technical features may be equivalently replaced; and the modifications or the substitutions do not make the essence of the corresponding technical solutions depart from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A liquid level data acquisition method comprising:

acquiring a detection image containing a liquid visible area;

and calculating the liquid level data according to the key point information.

2. The liquid level data acquisition method of claim 1, wherein the calculating the liquid level data from the keypoint information comprises:

acquiring the pixel coordinate of the middle point of the liquid level according to the pixel coordinates of the left point of the liquid level and the right point of the liquid level;

and calculating the liquid level data according to the proportion of the projection distances of the liquid level midpoint and the starting point and the ending point in the longitudinal axis direction of the liquid level visible region.

3. The liquid level data acquisition method according to claim 1, wherein the method further comprises:

and according to the key point information, when the coordinate value of the liquid level left side point or the liquid level right side point in the direction of the longitudinal axis is higher than the coordinate value of the end point in the direction of the longitudinal axis, or when the projection distance of the liquid level left side point and the liquid level right side point in the direction of the longitudinal axis is greater than a preset threshold value, determining that the detection image is invalid.

4. The liquid level data acquisition method according to claim 3,

the key point detection processing for the detection image includes:

a key point detection model is adopted to carry out key point detection processing on the detection image,

the method further comprises the following steps:

and taking the detection image determined to be invalid and the corresponding key point information thereof as training data, and training the key point detection model.

5. The liquid level data acquisition method according to claim 1, wherein when no liquid level object is detected in the detection image, the method further comprises:

performing liquid detection processing on the detection image by adopting a first deep learning classification model, and determining that the liquid level data is zero when no liquid object is detected in the detection image; when a liquid object is detected in the detection image, the liquid level data is determined to be a preset value.

6. The liquid level data acquisition method according to claim 1, wherein the acquiring of the detection image containing the visible region of the liquid comprises:

acquiring at least one frame of original image;

adopting a third deep learning classification model to carry out image quality detection processing on the original image;

when the image quality of the original image is detected to be higher than a preset threshold value, acquiring a meter type related to the liquid level data and coordinates of a detection frame comprising the liquid visible region by adopting a deep learning detection model;

and according to the coordinates of the detection frame, cutting the original image to obtain the detection image.

7. The liquid level data acquisition method according to claim 1, wherein the performing liquid level detection processing on the detection image includes:

and carrying out liquid level object detection processing on the detection image by adopting a second deep learning classification model.

8. A liquid level data acquisition interface for acquiring liquid level data of a liquid level type gauge, wherein the interface comprises:

the data display area is used for displaying the liquid level data;

9. The fluid level data acquisition interface of claim 8, wherein the image display area is further configured to display a detection frame identifying the visible region of fluid.

10. A computer-readable storage medium, on which a computer program is stored which is executable by a processor, wherein the program, when executed by the processor, implements the liquid level data acquisition method as claimed in any one of claims 1 to 7.