Abstract: This work aims at the automatic polyp detection by using a model of polyp appearance in the context of the analysis of colonoscopy videos. Our method consists of three stages: region segmentation, region description and region classification. The performance of our region segmentation method guarantees that if a polyp is present in the image, it will be exclusively and totally contained in a single region. The output of the algorithm also defines which regions can be considered as non-informative. We define as our region descriptor the novel Sector Accumulation-Depth of Valleys Accumulation (SA-DOVA), which provides a necessary but not sufficient condition for the polyp presence. Finally, we classify our segmented regions according to the maximal values of the SA-DOVA descriptor. Our preliminary classification results are promising, especially when classifying those parts of the image that do not contain a polyp inside.

Abstract: Research in computer vision is advancing by the availability of good datasets that help to improve algorithms, validate results and obtain comparative analysis. The datasets can be real or synthetic. For some of the computer vision problems such as optical flow it is not possible to obtain ground-truth optical flow with high accuracy in natural outdoor real scenarios directly by any sensor, although it is possible to obtain ground-truth data of real scenarios in a laboratory setup with limited motion. In this difficult situation computer graphics offers a viable option for creating realistic virtual scenarios. In the current work we present a framework to design virtual scenes and generate sequences as well as ground-truth flow fields. Particularly, we generate a dataset containing sequences of driving scenarios. The sequences in the dataset vary in different speeds of the on-board vision system, different road textures, complex motion of vehicle and independent moving vehicles in the scene. This dataset enables analyzing and adaptation of existing optical flow methods, and leads to invention of new approaches particularly for driver assistance systems.

Abstract: In this paper we describe a framework to learn a model of shape variability in a set of patterns. The framework is based on the Active Appearance Model (AAM) and permits to combine shape deformations with appearance variability. We have used two modifications of the Blurred Shape Model (BSM) descriptor as basic shape and appearance features to learn the model. These modifications permit to overcome the rigidity of the original BSM, adapting it to the deformations of the shape to be represented. We have applied this framework to representation and classification of handwritten digits and symbols. We show that results of the proposed methodology outperform the original BSM approach.

Abstract: Modern computer vision requires processing large amounts of data, both while training the model and/or during inference, once the model is deployed. Scenarios where images are captured and processed in physically separated locations are increasingly common (e.g. autonomous vehicles, cloud computing). In addition, many devices suffer from limited resources to store or transmit data (e.g. storage space, channel capacity). In these scenarios, lossy image compression plays a crucial role to effectively increase the number of images collected under such constraints. However, lossy compression entails some undesired degradation of the data that may harm the performance of the downstream analysis task at hand, since important semantic information may be lost in the process. Moreover, we may only have compressed images at training time but are able to use original images at inference time, or vice versa, and in such a case, the downstream model suffers from covariate shift. In this paper, we analyze this phenomenon, with a special focus on vision-based perception for autonomous driving as a paradigmatic scenario. We see that loss of semantic information and covariate shift do indeed exist, resulting in a drop in performance that depends on the compression rate. In order to address the problem, we propose dataset restoration, based on image restoration with generative adversarial networks (GANs). Our method is agnostic to both the particular image compression method and the downstream task; and has the advantage of not adding additional cost to the deployed models, which is particularly important in resource-limited devices. The presented experiments focus on semantic segmentation as a challenging use case, cover a broad range of compression rates and diverse datasets, and show how our method is able to significantly alleviate the negative effects of compression on the downstream visual task.

Abstract: In this paper, we extend the standard belief propagation (BP) sequential technique proposed in the tree-reweighted sequential method [15] to the fully connected CRF models with the geodesic distance affinity. The proposed method has been applied to the stereo matching problem. Also a new approach to the BP marginal solution is proposed that we call one-view occlusion detection (OVOD). In contrast to the standard winner takes all (WTA) estimation, the proposed OVOD solution allows to find occluded regions in the disparity map and simultaneously improve the matching result. As a result we can perform only
one energy minimization process and avoid the cost calculation for the second view and the left-right check procedure. We show that the OVOD approach considerably improves results for cost augmentation and energy minimization techniques in comparison with the standard one-view affinity space implementation. We apply our method to the Middlebury data set and reach state-ofthe-art especially for median, average and mean squared error metrics.

Abstract: In this paper, we aim to reconstruct the 3D motion parameters of a human body
model from the known 2D positions of a reduced set of joints in the image plane.
Towards this end, an action-specific motion model is trained from a database of real
motion-captured performances. The learnt motion model is used within a particle
filtering framework as a priori knowledge on human motion. First, our dynamic
model guides the particles according to similar situations previously learnt. Then, the solution space is constrained so only feasible human postures are accepted as valid solutions at each time step. As a result, we are able to track the 3D configuration of the full human body from several cycles of walking motion sequences using only the 2D positions of a very reduced set of joints from lateral or frontal viewpoints.

Abstract: This paper addresses the problem of inferring unseen cross-modal image-to-image translations between multiple modalities. We assume that only some of the pairwise translations have been seen (i.e. trained) and infer the remaining unseen translations (where training pairs are not available). We propose mix and match networks, an approach where multiple encoders and decoders are aligned in such a way that the desired translation can be obtained by simply cascading the source encoder and the target decoder, even when they have not interacted during the training stage (i.e. unseen). The main challenge lies in the alignment of the latent representations at the bottlenecks of encoder-decoder pairs. We propose an architecture with several tools to encourage alignment, including autoencoders and robust side information and latent consistency losses. We show the benefits of our approach in terms of effectiveness and scalability compared with other pairwise image-to-image translation approaches. We also propose zero-pair cross-modal image translation, a challenging setting where the objective is inferring semantic segmentation from depth (and vice-versa) without explicit segmentation-depth pairs, and only from two (disjoint) segmentation-RGB and depth-RGB training sets. We observe that a certain part of the shared information between unseen modalities might not be reachable, so we further propose a variant that leverages pseudo-pairs which allows us to exploit this shared information between the unseen modalities

Abstract: Two key factors affecting the performance of Error Correcting Output Codes (ECOC) in multiclass classification problems are the independence of binary classifiers and the problem-dependent coding design. In this paper, we propose an evolutionary algorithm-based approach to the design of an application-dependent codematrix in the ECOC framework. The central idea of this work is to design a three-dimensional codematrix, where the third dimension is the feature space of the problem domain. In order to do that, we consider the feature space in the design process of the codematrix with the aim of improving the independence and accuracy of binary classifiers. The proposed method takes advantage of some basic concepts of ensemble classification, such as diversity of classifiers, and also benefits from the evolutionary approach for optimizing the three-dimensional codematrix, taking into account the problem domain. We provide a set of experimental results using a set of benchmark datasets from the UCI Machine Learning Repository, as well as two real multiclass Computer Vision problems. Both sets of experiments are conducted using two different base learners: Neural Networks and Decision Trees. The results show that the proposed method increases the classification accuracy in comparison with the state-of-the-art ECOC coding techniques.