CN115482445A - Coal mine well wall water seepage identification method based on deep neural network - Google Patents
Coal mine well wall water seepage identification method based on deep neural network Download PDFInfo
- Publication number
- CN115482445A CN115482445A CN202211237913.6A CN202211237913A CN115482445A CN 115482445 A CN115482445 A CN 115482445A CN 202211237913 A CN202211237913 A CN 202211237913A CN 115482445 A CN115482445 A CN 115482445A
- Authority
- CN
- China
- Prior art keywords
- coal mine
- water seepage
- well wall
- image
- mine well
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 239000003245 coal Substances 0.000 title claims abstract description 73
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 title claims abstract description 64
- 238000000034 method Methods 0.000 title claims abstract description 41
- 238000013528 artificial neural network Methods 0.000 title claims abstract description 24
- 238000001514 detection method Methods 0.000 claims abstract description 26
- 230000007246 mechanism Effects 0.000 claims abstract description 20
- 230000003993 interaction Effects 0.000 claims abstract description 12
- 238000012549 training Methods 0.000 claims abstract description 8
- 230000004927 fusion Effects 0.000 claims abstract description 7
- 238000007781 pre-processing Methods 0.000 claims abstract description 4
- 238000010606 normalization Methods 0.000 claims description 15
- 238000013507 mapping Methods 0.000 claims description 13
- 230000008569 process Effects 0.000 claims description 13
- 238000010586 diagram Methods 0.000 claims description 8
- 238000002372 labelling Methods 0.000 claims description 4
- 238000012795 verification Methods 0.000 claims description 4
- 239000011159 matrix material Substances 0.000 claims description 3
- 230000009467 reduction Effects 0.000 claims description 3
- 238000013519 translation Methods 0.000 claims description 3
- 230000017105 transposition Effects 0.000 claims description 3
- 230000008595 infiltration Effects 0.000 claims 2
- 238000001764 infiltration Methods 0.000 claims 2
- 230000007306 turnover Effects 0.000 claims 2
- 238000013527 convolutional neural network Methods 0.000 description 4
- 230000007547 defect Effects 0.000 description 3
- 238000013135 deep learning Methods 0.000 description 2
- 238000007689 inspection Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000012512 characterization method Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 238000005065 mining Methods 0.000 description 1
- 238000003062 neural network model Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
- G06V10/80—Fusion, i.e. combining data from various sources at the sensor level, preprocessing level, feature extraction level or classification level
- G06V10/806—Fusion, i.e. combining data from various sources at the sensor level, preprocessing level, feature extraction level or classification level of extracted features
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Evolutionary Computation (AREA)
- Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Computing Systems (AREA)
- Software Systems (AREA)
- Multimedia (AREA)
- Medical Informatics (AREA)
- Databases & Information Systems (AREA)
- Biomedical Technology (AREA)
- Mathematical Physics (AREA)
- General Engineering & Computer Science (AREA)
- Molecular Biology (AREA)
- Data Mining & Analysis (AREA)
- Computational Linguistics (AREA)
- Biophysics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Geophysics And Detection Of Objects (AREA)
Abstract
The invention discloses a coal mine well wall water seepage identification method based on a deep neural network, which comprises the steps of obtaining a coal mine well wall image by utilizing image acquisition equipment, and selecting an image containing water seepage as an original coal mine well wall water seepage image; preprocessing an original coal mine well wall water seepage image to construct a coal mine well wall water seepage detection data set; constructing an improved YOLOv5 model; inputting a coal mine well wall seepage detection data set into an improved YOLOv5 model for training; and inputting the image to be detected into the trained model to complete the detection of the water seepage target. A Cross-Transformer module is added in a neck network of an original YOLOv5 model of the improved YOLOv5 model, a multi-head Cross attention mechanism is adopted in the Cross-Transformer module, interaction and fusion of key information among different levels of feature maps are achieved, the problem of long-distance dependence is solved, and the detection precision is improved from 78% to 85%.
Description
Technical Field
The invention relates to the technical field of coal mine well wall image detection. In particular to a coal mine well wall seepage identification method based on a deep neural network.
Background
Various huge losses are usually brought by water disasters in the coal mining process, water seepage detection is carried out on the coal mine well wall at regular intervals, and larger property losses caused by water seepage can be eliminated in time and the life safety of coal mine workers is protected.
At present, a coal mine well wall water seepage detection method is single, most of the methods adopt a manual inspection mode, but the manual inspection mode has obvious defects in the aspects of accuracy, efficiency, cost and the like. In recent years, the field of deep learning represented by a convolutional neural network has been rapidly developed, and a method based on deep learning has become an important means in the field of image detection.
At present, a plurality of neural network detection methods can be adopted for image detection, for example, a current popular YOLO series network model, but the YOLO series network model is created based on a convolutional neural network, lacks effective utilization of spatial position information, and cannot solve the problem of long-distance dependence. The neural network model comprises a plurality of convolution layers in the characteristic extraction process, and characteristic graphs of different levels can be generated. Wherein the low-level feature map contains more detailed information and location information, such as color, texture, etc. The high-level feature map contains more semantic information. Most of the existing methods lack information interaction among different hierarchical features, information is easily lost, and the detection accuracy is reduced.
Disclosure of Invention
Therefore, the technical problem to be solved by the invention is to provide a coal mine well wall water seepage identification method based on a deep neural network, which makes up the defects of a convolutional neural network and improves the detection precision of the water seepage target under the complex coal mine well wall environment.
In order to solve the technical problems, the invention provides the following technical scheme:
a coal mine well wall seepage water identification method based on a deep neural network,
step 100: acquiring a coal mine well wall image by using image acquisition equipment, and selecting an image containing a clear water seepage area as an original coal mine well wall water seepage image;
step 200: preprocessing an original coal mine well wall water seepage image to construct a coal mine well wall water seepage detection data set;
step 300: constructing an improved YOLOv5 model, and adding a Cross-Transformer module in an original neck network of the YOLOv5 model, wherein a multi-head Cross attention mechanism is adopted in the Cross-Transformer module;
step 400: inputting a coal mine well wall seepage detection data set into an improved YOLOv5 model for training;
step 500: and inputting the image to be detected into the trained model to complete the detection of the water seepage target.
In the coal mine well wall seepage identification method based on the deep neural network, in step 100, the number of the seepage-containing images is selected to be more than or equal to 300.
In the coal mine well wall water seepage identification method based on the deep neural network, in the step 200, a water seepage detection data set is established, and the method comprises the following substeps:
step 210: manually marking a water seepage area in an original coal mine well wall water seepage image by using LabelImg image marking software so as to form a coal mine well wall water seepage image data set, wherein a sample in the coal mine well wall water seepage image data set comprises a water seepage image and a water seepage target marking result graph corresponding to the water seepage image;
step 220: carrying out random rotation, translation, horizontal turning and vertical turning on images in the coal mine well wall seepage image data set to realize data expansion, and obtaining a coal mine well wall seepage image data set with the quantity expanded by four times, wherein the coal mine well wall seepage image data set contains more than or equal to 1200 seepage images;
step 230: and randomly dividing the samples in the expanded coal mine well wall seepage image data set into a training set and a verification set according to a preset proportion 8:2.
In step 300, the multi-head cross attention mechanism adopts a dual-input structure, and aims to obtain a correlation between an input feature map query and a key, and then multiplies another input value to obtain an attention map, and this process can be described as follows:
in the formula (1), Q 1 Querying Query, K, for an input feature graph 1 Is an input Key, V 1 Value, which is an input Value; q 2 Query for another input feature graph, K 2 Key of another input, V 2 Value, another input; attention stands for Attention mechanism algorithm;
C k t is the dimension of input data, and T is matrix transposition operation;
MCA 1 and MCA 2 Indicating the multi-headed cross attention corresponding to the two inputs.
According to the coal mine well wall water seepage identification method based on the deep neural network, the Cross-Transformer module adopts a double-branch structure, wherein the double-branch structure comprises linear projection, a feedforward network and layer normalization operation, and information interaction between two branches is realized by adopting a multi-head Cross attention mechanism.
According to the coal mine well wall water seepage identification method based on the deep neural network, two input features are mapped to x 1 And x 2 Performing dimension reduction operation, introducing learnable position embedding for coding position information of the feature mapping, respectively fusing the learnable position embedding with the input feature mapping in a direct adding mode, and then performing y 1 And y 2 Carrying out layer normalization operation to obtain the characteristic y after characteristic embedding 1 And y 2 This process can be described as:
in equation (2), LP is linear projection operation, PE is position coding, and LN represents slice normalization.
The coal mine well wall water seepage identification method based on the deep neural network realizes y by using a multi-head cross attention mechanism 1 And y 2 Information interaction between the characteristic graphs is carried out and added with the original characteristic graph to obtain a characteristic graph z 1 And z 2 This process can be described as:
in the formula, MCA 1 And MCA 2 Respectively representing two inputs y 1 And y 2 Corresponding multi-headed cross attention.
In order to adapt to the next dimension of the network, the coal mine well wall water seepage identification method based on the deep neural network adopts layer normalization, feed-forward network and feature mapping operation, and two feature maps z 1 And z 2 Adding to obtain a characteristic diagram f to realize further information fusion, wherein the process can be described as;
f＝FM[LN(z 1 )+LN(FFN(z 1 ))]+FM[LN(z 2 )+LN(FFN(z 2 ))] (3)；
in equation (3), LN represents layer normalization, FM represents feature mapping, and FFN is a feed-forward network.
The technical scheme of the invention achieves the following beneficial technical effects:
the application provides a coal mine well wall water seepage identification method based on a deep neural network, wherein a Cross-Transformer module is introduced into a neck network of an original YOLOv5 model, so that the defects of a convolutional neural network are overcome, and the problem of modeling remote dependence is effectively solved.
In addition, the invention also uses a multi-head Cross attention mechanism in a Cross-Transformer module to realize information interaction between different levels of characteristic graphs. Information interaction and fusion among feature graphs with the same scale and different levels in a Yolov5 model neck network are realized, and the characterization capability of the network is improved.
The improved YOLOv5 model is characterized in that a Cross-Transformer module is added in a neck network of an original YOLOv5 model, a multi-head Cross attention mechanism is adopted in the Cross-Transformer module, interaction and fusion of key information among different hierarchical characteristic diagrams can be realized, the problem of spatial long-distance dependence is solved, and the detection precision of a seepage target in a complex coal mine well wall environment is improved from 78% to 85%.
Drawings
FIG. 1 is a flow chart of a coal mine well wall seepage identification method based on a deep neural network in the invention;
FIG. 2 is a diagram of the existing YOLOv5 model framework;
FIG. 3 is a diagram of the improved YOLOv5 model framework in the present invention;
FIG. 4 is a schematic diagram of the Cross-Transfromer module structure in the present invention.
Detailed Description
Referring to fig. 1, the invention provides a coal mine well wall water seepage identification method based on a deep neural network, which is used for coal mine well wall water seepage detection work of a coal mine underground robot.
Step 100: acquiring a coal mine well wall image by using image acquisition equipment, and selecting an image containing water seepage as an original coal mine well wall water seepage image; obtaining 300 coal mine well wall seepage images;
step 200: preprocessing an original coal mine well wall water seepage image to construct a coal mine well wall water seepage detection data set;
step 210: manually labeling a water seepage area in an original coal mine well wall water seepage image by using LabelImg image labeling software so as to form a coal mine well wall water seepage image dataset, wherein a sample in the coal mine well wall water seepage image dataset comprises a water seepage image and a water seepage target labeling result image corresponding to the water seepage image;
step 220: carrying out random rotation, translation, horizontal turning and vertical turning operations on the images in the coal mine well wall seepage image data set to realize data expansion, and obtaining an expanded coal mine well wall seepage image data set; containing 1200 water seepage images;
step 230: and randomly dividing samples in the expanded coal mine well wall seepage image data set into a training set and a verification set according to the proportion of 8:2. The number of training set samples is 1080, and the number of verification set samples is 120.
Step 300: constructing an improved YOLOv5 model, and adding a Cross-Transformer module in an original neck network of the YOLOv5 model, wherein a multi-head Cross attention mechanism is adopted in the Cross-Transformer module;
the multi-head cross attention mechanism adopts a dual-input structure, the aim of the mechanism is to obtain the correlation between one input feature map query and a key, and then the correlation is multiplied by the value of another input to obtain an attention map, and the process can be described as follows:
in the formula (1), Q 1 Querying Query, K, for an input feature graph 1 Is an input Key, V 1 Value, which is an input Value; q 2 Query for another input feature graph, K 2 Key of another input, V 2 Value, another input; attention stands for Attention mechanism algorithm;
C k is the dimensionality of the input data; t is matrix transposition operation;
MCA 1 and MCA 2 Indicating the multi-headed cross attention corresponding to the two inputs.
As shown in fig. 4, the Cross-Transformer module adopts a dual-branch structure, which includes linear projection, feed-forward network, layer normalization operation, and realizes information interaction between two branches by using a multi-head Cross attention mechanism.
Step 310: mapping x two features of an input 1 And x 2 Performing dimension reduction operation, introducing learnable position embedding for coding position information of the feature mapping, respectively fusing the learnable position embedding with the input feature mapping in a direct adding mode, and then performing y 1 And y 2 Carrying out layer normalization operation to obtain the characteristic y after characteristic embedding 1 And y 2 This process can be described as:
in equation (2), LP is linear projection operation, PE is position coding, and LN represents slice normalization.
Step 320: by usingImplementation of multi-head cross attention mechanism 1 And y 2 Information interaction between the characteristic graphs is carried out and added with the original characteristic graph to obtain a characteristic graph z 1 And z 2 This process can be described as:
in the formula, MCA 1 And MCA 2 Respectively representing two inputs y 1 And y 2 Corresponding multi-headed cross attention.
Step 330: in order to adapt to the next dimension of the network, layer normalization, feed-forward network, feature mapping operation are adopted, and two feature maps z are combined 1 And z 2 Adding to obtain a characteristic diagram f to realize further information fusion, wherein the process can be described as;
f＝FM[LN(z 1 )+LN(FFN(z 1 ))]+FM[LN(z 2 )+LN(FFN(z 2 ))] (3)；
in equation (3), LN represents layer normalization, FM represents feature mapping, and FFN is a feed-forward network.
Step 400: inputting a coal mine well wall seepage detection data set into an improved YOLOv5 model for training;
step 500: and inputting the image to be detected into the trained model to complete the detection of the water seepage target.
The improved YOLOv5 model is characterized in that a Cross-Transformer module is added in a neck network of an original YOLOv5 model, a multi-head Cross attention mechanism is adopted in the Cross-Transformer module, interaction and fusion of key information among different hierarchical characteristic diagrams can be realized, the problem of spatial long-distance dependence is solved, and the detection precision of a seepage target in a complex coal mine well wall environment is improved from 78% to 85%.
It should be understood that the above examples are only for clarity of illustration and are not intended to limit the embodiments. Other variations and modifications will be apparent to persons skilled in the art in light of the above description. And are neither required nor exhaustive of all embodiments. And obvious variations or modifications derived therefrom are intended to be within the scope of the claims of this patent.
Claims (8)
1. A coal mine well wall water seepage identification method based on a deep neural network is characterized in that,
step 100: acquiring a coal mine well wall image by using image acquisition equipment, and selecting an image containing a clear water seepage area as an original coal mine well wall water seepage image;
step 200: preprocessing an original coal mine well wall water seepage image to construct a coal mine well wall water seepage detection data set;
step 300: constructing an improved YOLOv5 model, and adding a Cross-Transformer module in an original neck network of the YOLOv5 model, wherein a multi-head Cross attention mechanism is adopted in the Cross-Transformer module;
step 400: inputting a coal mine well wall seepage detection data set into an improved YOLOv5 model for training;
step 500: and inputting the image to be detected into the trained model to complete the detection of the water seepage target.
2. The coal mine borehole wall infiltration identification method based on the deep neural network as claimed in claim 1, wherein in step 100, the number of images containing infiltration is selected to be greater than or equal to 300.
3. The coal mine borehole wall water seepage identification method based on the deep neural network as claimed in claim 1, wherein in step 200, the water seepage detection data set is established, and the method comprises the following substeps:
step 210: manually labeling a water seepage area in an original coal mine well wall water seepage image by using LabelImg image labeling software so as to form a coal mine well wall water seepage image dataset, wherein a sample in the coal mine well wall water seepage image dataset comprises a water seepage image and a water seepage target labeling result image corresponding to the water seepage image;
step 220: carrying out random rotation, translation, horizontal turnover and vertical turnover on images in the coal mine well wall seepage image data set to realize data expansion, and obtaining a coal mine well wall seepage image data set with the quantity expanded by four times, wherein the seepage images are larger than or equal to 1200;
step 230: and randomly dividing samples in the expanded coal mine well wall seepage image data set into a training set and a verification set according to a preset proportion of 8:2.
4. The method for identifying coal mine borehole wall seepage based on the deep neural network as claimed in claim 1, wherein in step 300, the multi-head cross attention mechanism adopts a dual-input structure, the purpose of which is to obtain the correlation between one input feature map query and a key, and then to multiply with another input value to obtain an attention map, and this process can be described as follows:
in the formula (1), Q 1 Querying Query, K, for an input feature graph 1 Key, V, for an input 1 Value, which is an input Value; q 2 Query for another input feature graph, K 2 Key of another input, V 2 Value, another input; attention stands for Attention mechanism algorithm;
C k t is the dimension of input data and is the matrix transposition operation;
MCA 1 and MCA 2 Indicating the multi-headed cross attention corresponding to the two inputs.
5. The coal mine borehole wall seepage identification method based on the deep neural network is characterized in that the Cross-Transformer module adopts a double-branch structure, wherein the double-branch structure comprises linear projection, a feed-forward network and layer normalization operation, and a multi-head Cross attention mechanism is adopted to realize information interaction between two branches.
6. The coal mine borehole wall seepage identification method based on the deep neural network as claimed in claim 5, wherein x is mapped to two input features 1 And x 2 Performing dimension reduction operation, introducing learnable position embedding for coding position information of the feature mapping, respectively fusing the learnable position embedding with the input feature mapping in a direct adding mode, and then carrying out y 1 And y 2 Carrying out layer normalization operation to obtain the feature y after embedding the feature 1 And y 2 This process can be described as:
in equation (2), LP is linear projection operation, PE is position coding, and LN represents slice normalization.
7. The coal mine well wall seepage identification method based on the deep neural network as claimed in claim 6, wherein y is realized by using a multi-head cross attention mechanism 1 And y 2 Information interaction between the characteristic graphs is carried out and added with the original characteristic graph to obtain a characteristic graph z 1 And z 2 This process can be described as:
in the formula, MCA 1 And MCA 2 Respectively representing two inputs y 1 And y 2 Corresponding multi-headed cross attention.
8. The coal mine borehole wall seepage identification method based on the deep neural network as claimed in claim 7, characterized in that in order to adapt to the next dimension of the network, layer normalization, feed-forward network and feature mapping operation are adopted, and two feature maps z are combined 1 And z 2 Adding to obtain a characteristic diagram f for further information fusion, whichA process may be described as;
f＝FM[LN(z 1 )+LN(FFN(z 1 ))]+FM[LN(z 2 )+LN(FFN(z 2 ))] (3)；
in equation (3), LN indicates layer normalization, FM indicates feature mapping, and FFN is a feed-forward network.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211237913.6A CN115482445A (en) | 2022-10-10 | 2022-10-10 | Coal mine well wall water seepage identification method based on deep neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211237913.6A CN115482445A (en) | 2022-10-10 | 2022-10-10 | Coal mine well wall water seepage identification method based on deep neural network |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115482445A true CN115482445A (en) | 2022-12-16 |
Family
ID=84393221
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211237913.6A Pending CN115482445A (en) | 2022-10-10 | 2022-10-10 | Coal mine well wall water seepage identification method based on deep neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115482445A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116542965A (en) * | 2023-06-07 | 2023-08-04 | 西安热工研究院有限公司 | A method, device, equipment, and readable storage medium for detecting water leakage in industrial pipelines |
| CN119399493A (en) * | 2024-10-17 | 2025-02-07 | 山西省信息产业技术研究院有限公司 | Coal gangue detection method based on multi-angle perception and mixed-scale Transformer feature aggregation |
| CN119516447A (en) * | 2025-01-20 | 2025-02-25 | 浙江浙能迈领环境科技股份有限公司 | A ship sewage well monitoring system and method |
-
2022
- 2022-10-10 CN CN202211237913.6A patent/CN115482445A/en active Pending
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116542965A (en) * | 2023-06-07 | 2023-08-04 | 西安热工研究院有限公司 | A method, device, equipment, and readable storage medium for detecting water leakage in industrial pipelines |
| CN119399493A (en) * | 2024-10-17 | 2025-02-07 | 山西省信息产业技术研究院有限公司 | Coal gangue detection method based on multi-angle perception and mixed-scale Transformer feature aggregation |
| CN119399493B (en) * | 2024-10-17 | 2025-04-15 | 山西省信息产业技术研究院有限公司 | Coal gangue detection method based on multi-angle perception and mixed-scale Transformer feature aggregation |
| CN119516447A (en) * | 2025-01-20 | 2025-02-25 | 浙江浙能迈领环境科技股份有限公司 | A ship sewage well monitoring system and method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN115482445A (en) | Coal mine well wall water seepage identification method based on deep neural network | |
| CN112035453B (en) | Recommendation method, device and storage medium based on GBDT high-order feature combination | |
| Denby | Discovering the importance of mesoscale cloud organization through unsupervised classification | |
| US11093576B2 (en) | Core-plug to giga-cells lithological modeling | |
| CN100373378C (en) | Data processing and information management method for submarine hydrothermal activity detection | |
| CN109684428A (en) | Spatial data building method, device, equipment and storage medium | |
| CN103984624A (en) | Automatic testing method and system for webpage | |
| CN113033403A (en) | Image tile-based ecological protection red line ground object target identification method and system | |
| Guo et al. | Adaln: a vision transformer for multidomain learning and predisaster building information extraction from images | |
| CN104866630B (en) | GIS and the automatic spatial modeling system and methods of ES in metallogenic prognosis | |
| CN116310628A (en) | Token mask mechanism-based large-scale village-in-city extraction method | |
| Peters et al. | A new biogeographic regionalisation for Tasmania | |
| Te Stroet et al. | Predicting sweet spots in shale plays by DNA fingerprinting and machine learning | |
| CN115497006B (en) | Urban remote sensing image change depth monitoring method and system based on dynamic mixing strategy | |
| Jiang et al. | Scan4Façade: Automated as-is façade modeling of historic high-rise buildings using drones and AI | |
| Ouyang et al. | MDFF: A method for fine-grained UFZ mapping with multimodal geographic data and deep network | |
| Galantucci et al. | Machine learning for the semi-automatic 3D decay segmentation and mapping of heritage assets | |
| CN117933717B (en) | Underground water pollution risk prevention and control area boundary accurate division method and system | |
| CN104008167A (en) | Geologic map database establishment model for describing geologic bodies through single feature | |
| Singh et al. | Uncertainty quantification of forecasted oil recovery using dynamic model ranking with application to a ME carbonate reservoir | |
| CN115393368A (en) | Method, system, equipment and medium for identifying colony site selection environment mode based on meta-learning | |
| Chinnari et al. | Semantic Segmentation in Aerial Imagery: A Novel Approach for Urban Planning and Development | |
| Chen et al. | Self-intersection attention pooling based classification for rock recognition | |
| CN120032269B (en) | A landslide intelligent identification model and method based on large model prior knowledge | |
| CN112182084A (en) | Data format-based data automatic processing method and equipment |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |