Abstract: Poster
paper TA.P5.12
Face detectors typically produce a large number of false positives and this leads to the need to have a further non maximum suppression stage to eliminate multiple and spurious responses. This stage is based on considering spatial heuristics: true positive responses are selected by implicitly considering several restrictions on the spatial distribution of detector responses in natural images. In this paper we analyze the limitations of this approach and propose an efficient search method to overcome them. Results show how the application of this new non-maximum suppression approach to a simple face detector boosts its performance to state of the art results.

Abstract: In this work, we address the study of intestinal contractions in a novel approach based on a machine learning framework to process data from Wireless Capsule Video Endoscopy. Wireless endoscopy represents a unique way to visualize the intestine motility by creating long videos to visualize intestine dynamics. In this paper we argue that to analyze huge amount of wireless endoscopy data and define robust methods for contraction detection we should base our approach on sophisticated machine learning techniques. In particular, we propose a cascade of classifiers in order to remove different physiological phenomenon and obtain the motility pattern of small intestines. Our results show obtaining high specificity and sensitivity rates that highlight the high efficiency of the selected approach and support the feasibility of the proposed methodology in the automatic detection and analysis of intestine contractions.

Abstract: Cross-modal retrieval methods have been significantly improved in last years with the use of deep neural networks and large-scale annotated datasets such as ImageNet and Places. However, collecting and annotating such datasets requires a tremendous amount of human effort and, besides, their annotations are limited to discrete sets of popular visual classes that may not be representative of the richer semantics found on large-scale cross-modal retrieval datasets. In this paper, we present a self-supervised cross-modal retrieval framework that leverages as training data the correlations between images and text on the entire set of Wikipedia articles. Our method consists in training a CNN to predict: (1) the semantic context of the article in which an image is more probable to appear as an illustration, and (2) the semantic context of its caption. Our experiments demonstrate that the proposed method is not only capable of learning discriminative visual representations for solving vision tasks like classification, but that the learned representations are better for cross-modal retrieval when compared to supervised pre-training of the network on the ImageNet dataset.

Abstract: The Multiple Appropriate Facial Reaction Generation Challenge (REACT2023) is the first competition event focused on evaluating multimedia processing and machine learning techniques for generating human-appropriate facial reactions in various dyadic interaction scenarios, with all participants competing strictly under the same conditions. The goal of the challenge is to provide the first benchmark test set for multi-modal information processing and to foster collaboration among the audio, visual, and audio-visual behaviour analysis and behaviour generation (a.k.a generative AI) communities, to compare the relative merits of the approaches to automatic appropriate facial reaction generation under different spontaneous dyadic interaction conditions. This paper presents: (i) the novelties, contributions and guidelines of the REACT2023 challenge; (ii) the dataset utilized in the challenge; and (iii) the performance of the baseline systems on the two proposed sub-challenges: Offline Multiple Appropriate Facial Reaction Generation and Online Multiple Appropriate Facial Reaction Generation, respectively. The challenge baseline code is publicly available at https://github.com/reactmultimodalchallenge/baseline_react2023.

Abstract: Dynamic tracking of heart moving is one relevant target in medical imag- ing and can be helpful for analyzing heart dynamics in the study of several cardiac diseases. For this aim, a previous segmentation problem of such structures is stated, based on certain relevant features (like edges or intensity levels, textures, etc.) Clas- sical active models have been used, but they fail when overlapping structures or not well-defined contours are present. Automatic feature learning systems may be a pow- erful tool. Discriminant active contours present optimal results in this kind of problem. They are a kind of deformable models that converge to an optimal object segmenta- tion that dynamically adapts to the object contour. The feature space is designed from a filter bank in order to guarantee the search and learning of the set of relevant fea- tures for optimal classification on each part of the object. Tracking of target evolution is obtained through the whole set of images, using information from the actual and previous stages. Feedback systems are implemented to guarantee the minimum well- separable classification set in each segmentation step. Our implementation has been proved with several series of Magnetic Resonance with improved results in segmenta- tion in comparison to previous methods.