Oscar Deniz

The phenomenon of Adversarial Examples has become one of the most intriguing topics associated to deep learning. The so-called adversarial attacks have the ability to fool deep neural networks with inappreciable perturbations. While the effect is striking, it has been suggested that such carefully selected injected noise does not necessarily appear in real-world scenarios. In contrast to this, some authors have looked for ways to generate adversarial noise in physical scenarios (traffic signs, shirts, etc.), thus showing that attackers can indeed fool the networks. In this paper we go beyond that and show that adversarial examples also appear in the real-world without any attacker or maliciously selected noise involved. We show this by using images from tasks related to microscopy and also general object recognition with the well-known ImageNet dataset. A comparison between these natural and the artificially generated adversarial examples is performed using distance metrics and imag...

One of the most intriguing phenomenons related to deep learning is the so-called adversarial examples. These samples are visually equivalent to normal inputs, undetectable for humans, yet they cause the networks to output wrong results. The phenomenon can be framed as a symmetry/asymmetry problem, whereby inputs to a neural network with a similar/symmetric appearance to regular images, produce an opposite/asymmetric output. Some researchers are focused on developing methods for generating adversarial examples, while others propose defense methods. In parallel, there is a growing interest in characterizing the phenomenon, which is also the focus of this paper. From some well known datasets of common images, like CIFAR-10 and STL-10, a neural network architecture is first trained in a normal regime, where training and validation performances increase, reaching generalization. Additionally, the same architectures and datasets are trained in an overfitting regime, where there is a growi...

Adversarial examples are one of the most intriguing topics in modern deep learning. Imperceptible perturbations to the input can fool robust models. In relation to this problem, attack and defense methods are being developed almost on a daily basis. In parallel, efforts are being made to simply pointing out when an input image is an adversarial example. This can help prevent potential issues, as the failure cases are easily recognizable by humans. The proposal in this work is to study how chaos theory methods can help distinguish adversarial examples from regular images. Our work is based on the assumption that deep networks behave as chaotic systems, and adversarial examples are the main manifestation of it (in the sense that a slight input variation produces a totally different output). In our experiments, we show that the Lyapunov exponents (an established measure of chaoticity), which have been recently proposed for classification of adversarial examples, are not robust to image...

An automatic “museum audio guide” is presented as a new type of audio guide for museums. The device consists of a headset equipped with a camera that captures exhibit pictures and the eyes of things computer vision device (EoT). The EoT board is capable of recognizing artworks using features from accelerated segment test (FAST) keypoints and a random forest classifier, and is able to be used for an entire day without the need to recharge the batteries. In addition, an application logic has been implemented, which allows for a special highly-efficient behavior upon recognition of the painting. Two different use case scenarios have been implemented. The main testing was performed with a piloting phase in a real world museum. Results show that the system keeps its promises regarding its main benefit, which is simplicity of use and the user’s preference of the proposed system over traditional audioguides.

Deep learning (henceforth DL) has become most powerful machine learning methodology. Under specific circumstances recognition rates even surpass those obtained by humans. Despite this, several works have shown that deep learning produces outputs that are very far from human responses when confronted with the same task. This the case of the so-called “adversarial examples” (henceforth AE). The fact that such implausible misclassifications exist points to a fundamental difference between machine and human learning. This paper focuses on the possible causes of this intriguing phenomenon. We first argue that the error in adversarial examples is caused by high bias, i.e. by regularization that has local negative effects. This idea is supported by our experiments in which the robustness to adversarial examples is measured with respect to the level of fitting to training samples. Higher fitting was associated to higher robustness to adversarial examples. This ties the phenomenon to the tra...

Computer vision and deep learning are clearly demonstrating a capability to create engaging cognitive applications and services. However, these applications have been mostly confined to powerful Graphic Processing Units (GPUs) or the cloud due to their demanding computational requirements. Cloud processing has obvious bandwidth, energy consumption and privacy issues. The Eyes of Things (EoT) is a powerful and versatile embedded computer vision platform which allows the user to develop artificial vision and deep learning applications that analyse images locally. In this article, we use the deep learning capabilities of an EoT device for a real-life facial informatics application: a doll capable of recognizing emotions, using deep learning techniques, and acting accordingly. The main impact and significance of the presented application is in showing that a toy can now do advanced processing locally, without the need of further computation in the cloud, thus reducing latency and removi...

While action recognition has become an important line of research in computer vision, the recognition of particular events such as aggressive behaviors, or fights, has been relatively less studied. These tasks may be extremely useful in several video surveillance scenarios such as psychiatric wards, prisons or even in personal camera smartphones. Their potential usability has led to a surge of interest in developing fight or violence detectors. One of the key aspects in this case is efficiency, that is, these methods should be computationally fast. "Handcrafted" spatiotemporal features that account for both motion and appearance information can achieve high accuracy rates, albeit the computational cost of extracting some of those features is still prohibitive for practical applications. The deep learning paradigm has been recently applied for the first time to this task too, in the form of a 3D Convolutional Neural Network that processes the whole video sequence as input. ...

We study the effectiveness of several low-cost oblique illumination filters to improve overall image quality, in comparison with standard bright field imaging. For this purpose, a dataset composed of 3360 diatom images belonging to 21 taxa was acquired. Subjective and objective image quality assessments were done. The subjective evaluation was performed by a group of diatom experts by psychophysical test where resolution, focus, and contrast were assessed. Moreover, some objective nonreference image quality metrics were applied to the same image dataset to complete the study, together with the calculation of several texture features to analyze the effect of these filters in terms of textural properties. Both image quality evaluation methods, subjective and objective, showed better results for images acquired using these illumination filters in comparison with the no filtered image. These promising results confirm that this kind of illumination filters can be a practical way to impro...

Application of deep learning algorithms to whole-slide pathology images can potentially improve diagnostic accuracy and efficiency. Assess the performance of automated deep learning algorithms at detecting metastases in hematoxylin and eosin-stained tissue sections of lymph nodes of women with breast cancer and compare it with pathologists' diagnoses in a diagnostic setting. Researcher challenge competition (CAMELYON16) to develop automated solutions for detecting lymph node metastases (November 2015-November 2016). A training data set of whole-slide images from 2 centers in the Netherlands with (n = 110) and without (n = 160) nodal metastases verified by immunohistochemical staining were provided to challenge participants to build algorithms. Algorithm performance was evaluated in an independent test set of 129 whole-slide images (49 with and 80 without metastases). The same test set of corresponding glass slides was also evaluated by a panel of 11 pathologists with time constr...

Immunohistochemical (IHC) biomarkers in breast tissue microarray (TMA) samples are used daily in pathology departments. In recent years, automatic methods to evaluate positive staining have been investigated since they may save time and reduce errors in the diagnosis. These errors are mostly due to subjective evaluation. The aim of this work is to develop a density tool able to automatically quantify the positive brown IHC stain in breast TMA for different biomarkers. To avoid the problem of colour variation and make a robust tool independent of the staining process, several colour standardization methods have been analysed. Four colour standardization methods have been compared against colour model segmentation. The standardization methods have been compared by means of NBS colour distance. The use of colour standardization helps to reduce noise due to stain and histological sample preparation. However, the most reliable and robust results have been obtained by combining the HSV an...

Breast cancer is the most common type of cancer and the fifth leading cause of death in women over 40. Therefore, prompt diagnostic and treatment is essential. In this work a TMA Computer Aided Diagnosis (CAD) system has been implemented to provide support to pathologists in their daily work. For that purpose, the tool covers each and every process from the TMA core image acquisition to their individual classification. The first process includes: tissue core location, segmentation and rigid registration of digital microscopic images acquired at different magnifications (5x, 10x, 20x, 20x and 40x) from different devices. The classification process allows performing the core classification selecting different types of color models, texture descriptors and classifiers. Finally, the cores are classified into three categories: malignant, doubtful and benign.

Advances in digital pathology are generating huge volumes of whole slide (WSI) and tissue microarray images (TMA) which are providing new insights into the causes of cancer. The challenge is to extract and process effectively all the information in order to characterize all the heterogeneous tissue-derived data. This study aims to identify an optimal set of features that best separates different classes in breast TMA. These classes are: stroma, adipose tissue, benign and benign anomalous structures and ductal and lobular carcinomas. To this end, we propose an exhaustive assessment on the utility of textons and colour for automatic classification of breast TMA. Frequential and spatial texton maps from eight different colour models were extracted and compared. Then, in a novel way, the TMA is characterized by the 1st and 2nd order Haralick statistical descriptors obtained from the texton maps with a total of 241×8 features for each original RGB image. Subsequently, a feature selection process is performed to remove redundant information and therefore to reduce the dimensionality of the feature vector. Three methods were evaluated: linear discriminant analysis, correlation and sequential forward search. Finally, an extended bank of classifiers composed of six techniques was compared, but only three of them could significantly improve accuracy rates: Fisher, Bagging Trees and AdaBoost. Our results reveal that the combination of different colour models applied to spatial texton maps provides the most efficient representation of the breast TMA. Specifically, we found that the best colour model combination is Hb, Luv and SCT for all classifiers and the classifier that performs best for all colour model combinations is the AdaBoost. On a database comprising 628 TMA images, classification yields an accuracy of 98.1% and a precision of 96.2% with a total of 316 features on spatial textons maps.

The field of anatomic pathology has experienced major changes over the last decade. Virtual microscopy (VM) systems have allowed experts in pathology and other biomedical areas to work in a safer and more collaborative way. VMs are automated systems capable of digitizing microscopic samples that were traditionally examined one by one. The possibility of having digital copies reduces the risk of damaging original samples, and also makes it easier to distribute copies among other pathologists. This article describes the development of an automated high-resolution whole slide imaging (WSI) system tailored to the needs and problems encountered in digital imaging for pathology, from hardware control to the full digitization of samples. The system has been built with an additional digital monochromatic camera together with the color camera by default and LED transmitted illumination (RGB). Monochrome cameras are the preferred method of acquisition for fluorescence microscopy. The system is able to digitize correctly and form large high resolution microscope images for both brightfield and fluorescence. The quality of the digital images has been quantified using three metrics based on sharpness, contrast and focus. It has been proved on 150 tissue samples of brain autopsies, prostate biopsies and lung cytologies, at five magnifications: 2.5×, 10×, 20×, 40×, and 63×. The article is focused on the hardware set-up and the acquisition software, although results of the implemented image processing techniques included in the software and applied to the different tissue samples are also presented.

This paper describes a general methodology for developing parallel image processing algorithms based on message passing for high resolution images (on the order of several Gigabytes). These algorithms have been applied to histological images and must be executed on massively parallel processing architectures. Advances in new technologies for complete slide digitalization in pathology have been combined with developments in biomedical informatics. However, the efficient use of these digital slide systems is still a challenge. The image processing that these slides are subject to is still limited both in terms of data processed and processing methods. The work presented here focuses on the need to design and develop parallel image processing tools capable of obtaining and analyzing the entire gamut of information included in digital slides. Tools have been developed to assist pathologists in image analysis and diagnosis, and they cover low and high-level image processing methods applied to histological images. Code portability, reusability and scalability have been tested by using the following parallel computing architectures: distributed memory with massive parallel processors and two networks, INFINIBAND and Myrinet, composed of 17 and 1024 nodes respectively. The parallel framework proposed is flexible, high performance solution and it shows that the efficient processing of digital microscopic images is possible and may offer important benefits to pathology laboratories.

ABSTRACT In face recognition, where high-dimensional representation spaces are generally used, it is very important to take advantage of all the available information. In particular, many labelled facial images will be accumulated while the recognition system is functioning, and due to practical reasons some of them are often discarded. In this paper, we propose an algorithm for using this information. The algorithm has the fundamental characteristic of being incremental. On the other hand, the algorithm makes use of a combination of classification results for the images in the input sequence. Experiments with sequences obtained with a real person detection and tracking system allow us to analyze the performance of the algorithm, as well as its potential improvements. Keywords: face recognition, incremental learning, face sequences 1

The future paradigm of pathology will be digital. Instead of conventional microscopy, a pathologist will perform a diagnosis through interacting with images on computer screens and performing quantitative analysis. The fourth generation of virtual slide telepathology systems, so-called virtual microscopy and whole-slide imaging (WSI), has allowed for the storage and fast dissemination of image data in pathology and other biomedical areas. These novel digital imaging modalities encompass high-resolution scanning of tissue slides and derived technologies, including automatic digitization and computational processing of whole microscopic slides. Moreover, automated image analysis with WSI can extract specific diagnostic features of diseases and quantify individual components of these features to support diagnoses and provide informative clinical measures of disease. Therefore, the challenge is to apply information technology and image analysis methods to exploit the new and emerging digital pathology technologies effectively in order to process and model all the data and information contained in WSI. The final objective is to support the complex workflow from specimen receipt to anatomic pathology report transmission, that is, to improve diagnosis both in terms of pathologists' efficiency and with new information. This article reviews the main concerns about and novel methods of digital pathology discussed at the latest workshop in the field carried out within the European project AIDPATH (Academia and Industry Collaboration for Digital Pathology).

... 2. J. Bouguet. Pyramidal implementation of the Lucas Kanade feature tracker. Technical report, Intel Corporation, Microprocessor Research Labs, OpenCV documents, 1999. 3. Cynthia Breazeal, Designing Sociable Robots, MIT Press, Cambridge, MA, 2002. 4. M. Castrillón. ...

An emergent field of robotics, which we may call social robotics, aims at building robots that have abilities to interact with people. These robots have expressive power (i.e. they have an expressive face, voice, etc.) as well as abilities to locate, pay attention to, and address people. For this class of robots, the dominant design approach has been that of following models taken from human sciences like developmental psychology, ethology and even neurophysiology. At the same time the engineering point of view as a main guideline has received less attention, which has not happened for other types of machines like industrial robots. We formally show that the reproduction of social intelligence, as opposed to other types of human abilities, leads to fragile performance, in the sense of having very different performances between training and working (unseen) conditions. The engineering point of view has therefore to be seriously considered if useful social robots are to be built. This...

Fast and reliable face and facial feature detection are required abilities for any Human Computer Interaction approach based on Computer Vision. Since the publication of the Viola-Jones object detection framework and the more recent open source implementation, an increasing number of applications have appeared, particularly in the context of facial processing. In this respect, the OpenCV community shares a collection of public domain classifiers for this scenario. However, as far as we know these classifiers have never been evaluated and/or compared. In this paper we analyze the individual performance of all those public classifiers getting the best performance for each target. These results are valid to define a baseline for future approaches. Additionally we propose a simple hierarchical combination of those classifiers to increase the facial feature detection rate while reducing the face false detection rate.

Abstract This paper does not propose a new technique for face representation or classification. Instead the work described here investigates the evolution of an automatic,system which, based on a currently common framework, and start- ing from an empty memory, modifies its classifiers accord- ing to experience. In the experiments we reproduce up to a certain extent the process of successive meetings. The results achieved, even when the number of different indi- viduals is still reduced compared to off-line classifiers, are promising.