Abstract: We present in this paper a novel study aiming at identifying the differences in visual search patterns between physicians of diverse levels of expertise during the screening of colonoscopy videos. Physicians were clustered into two groups -experts and novices- according to the number of procedures performed, and fixations were captured by an eye-tracker device during the task of polyp search in different video sequences. These fixations were integrated into heat maps, one for each cluster. The obtained maps were validated over a ground truth consisting of a mask of the polyp, and the comparison between experts and novices was performed by using metrics such as reaction time, dwelling time and energy concentration ratio. Experimental results show a statistically significant difference between experts and novices, and the obtained maps show to be a useful tool for the characterisation of the behaviour of each group.

Abstract: A common technique to avoid overfitting when training deep neural networks (DNN) is to monitor the performance in a dedicated validation data partition and to stop
training as soon as it saturates. This only focuses on what the model does, while completely ignoring what happens inside it.
In this work, we open the “black-box” of DNN in order to perform early stopping. We propose to use a novel theoretical framework that analyses meso-scale patterns in the topology of the functional graph of a network while it trains. Based on it,
we decide when it transitions from learning towards overfitting in a more explainable way. We exemplify the benefits of this approach on a state-of-the art custom DNN that jointly learns local representations and label structure employing an ensemble of dedicated subnetworks. We show that it is practically equivalent in performance to early stopping with patience, the standard early stopping algorithm in the literature. This proves beneficial for AU recognition performance and provides new insights into how learning of AUs occurs in DNNs.

Abstract: Person re-identification has received special attentionby the human analysis community in the last few years.To address the challenges in this field, many researchers haveproposed different strategies, which basically exploit eithercross-view invariant features or cross-view robust metrics. Inthis work we propose to combine different feature representationsthrough ranking aggregation. Spatial information, whichpotentially benefits the person matching, is represented usinga 2D body model, from which color and texture informationare extracted and combined. We also consider contextualinformation (background and foreground data), automaticallyextracted via Deep Decompositional Network, and the usage ofConvolutional Neural Network (CNN) features. To describe thematching between images we use the polynomial feature map,also taking into account local and global information. Finally,the Stuart ranking aggregation method is employed to combinecomplementary ranking lists obtained from different featurerepresentations. Experimental results demonstrated that weimprove the state-of-the-art on VIPeR and PRID450s datasets,achieving 58.77% and 71.56% on top-1 rank recognitionrate, respectively, as well as obtaining competitive results onCUHK01 dataset.

Abstract: Multi-domain recommender systems benefit from cross-domain representation learning and positive knowledge transfer. Both can be achieved by introducing a specific modeling of input data (i.e. disjoint history) or trying dedicated training regimes. At the same time, treating domains as separate input sources becomes a limitation as it does not capture the interplay that naturally exists between domains. In this work, we efficiently learn multi-domain representation of sequential users’ interactions using graph neural networks. We use temporal intra- and inter-domain interactions as contextual information for our method called MAGRec (short for Multi-dom Ain Graph-based Recommender). To better capture all relations in a multi-domain setting, we learn two graph-based sequential representations simultaneously: domain-guided for recent user interest, and general for long-term interest. This approach helps to mitigate the negative knowledge transfer problem from multiple domains and improve overall representation. We perform experiments on publicly available datasets in different scenarios where MAGRec consistently outperforms state-of-the-art methods. Furthermore, we provide an ablation study and discuss further extensions of our method.

Abstract: In this paper we present an example of a video surveillance application that exploits Multimodal Interactive (MI) technologies. The main objective of the so-called VID-Hum prototype was to develop a cognitive artificial system for both the detection and description of a particular set of human behaviours arising from real-world events. The main procedure of the prototype described in this chapter entails: (i) adaptation, since the system adapts itself to the most common behaviours (qualitative data) inferred from tracking (quantitative data) thus being able to recognize abnormal behaviors; (ii) feedback, since an advanced interface based on Natural Language understanding allows end-users the communicationwith the prototype by means of conceptual sentences; and (iii) multimodality, since a virtual avatar has been designed to describe what is happening in the scene, based on those textual interpretations generated by the prototype. Thus, the MI methodology has provided an adequate framework for all these cooperating processes.

Abstract: This thesis addresses the problem of automatic scene text understanding in unconstrained conditions. In particular, we tackle the tasks of multi-language and arbitrary-oriented text detection, tracking, and script identification in natural scenes.
For this we have developed a set of generic methods that build on top of the basic observation that text has always certain key visual and structural characteristics that are independent of the language or script in which it is written. Text instances in any
language or script are always formed as groups of similar atomic parts, being them either individual characters, small stroke parts, or even whole words in the case of cursive text. This holistic (sumof-parts) and recursive perspective has lead us to explore different variants of the “segmentation and grouping” paradigm of computer vision.
Scene text detection methodologies are usually based in classification of individual regions or patches, using a priory knowledge for a given script or language. Human perception of text, on the other hand, is based on perceptual organization through which
text emerges as a perceptually significant group of atomic objects.
In this thesis, we argue that the text detection problem must be posed as the detection of meaningful groups of regions. We address the problem of text detection in natural scenes from a hierarchical perspective, making explicit use of the recursive nature of text, aiming directly to the detection of region groupings corresponding to text within a hierarchy produced by an agglomerative similarity clustering process over individual regions. We propose an optimal way to construct such an hierarchy introducing a feature space designed to produce text group hypothese with high recall and a novel stopping rule combining a discriminative classifier and a probabilistic measure of group meaningfulness based in perceptual organization. Within this generic framework, we design a text-specific object proposals algorithm that, contrary to existing generic object proposals methods, aims directly to the detection of text regions groupings. For this, we abandon the rigid definition of “what is text” of traditional specialized text detectors, and move towards more fuzzy perspective of grouping-based object proposals methods.
Then, we present a hybrid algorithm for detection and tracking of scene text where the notion of region groupings plays also a central role. By leveraging the structural arrangement of text group components between consecutive frames we can improve
the overall tracking performance of the system.
Finally, since our generic detection framework is inherently designed for multi-language environments, we focus on the problem of script identification in order to build a multi-language end-toend reading system. Facing this problem with state of the art CNN classifiers is not straightforward, as they fail to address a key
characteristic of scene text instances: their extremely variable aspect ratio. Instead of resizing input images to a fixed size as in the typical use of holistic CNN classifiers, we propose a patch-based classification framework in order to preserve discriminative parts of the image that are characteristic of its class. We describe a novel method based on the use of ensembles of conjoined networks to jointly learn discriminative stroke-parts representations and their relative importance in a patch-based classification scheme.