Abstract: Deep learning has recently been enjoying an increasing popularity due to its success in solving challenging tasks. In particular, deep learning has proven to be effective in a large variety of computer vision tasks, such as image classification, object recognition and image parsing. Contrary to previous research, which required engineered feature representations, designed by experts, in order to succeed, deep learning attempts to learn representation hierarchies automatically from data. More recently, the trend has been to go deeper with representation hierarchies.
Learning (very) deep representation hierarchies is a challenging task, which
involves the optimization of highly non-convex functions. Therefore, the search
for algorithms to ease the learning of (very) deep representation hierarchies from data is extensive and ongoing.
In this thesis, we tackle the challenging problem of easing the learning of (very) deep representation hierarchies. We present a hyper-parameter free, off-the-shelf, simple and fast unsupervised algorithm to discover hidden structure from the input data by enforcing a very strong form of sparsity. We study the applicability and potential of the algorithm to learn representations of varying depth in a handful of applications and domains, highlighting the ability of the algorithm to provide discriminative feature representations that are able to achieve top performance.
Yet, while emphasizing the great value of unsupervised learning methods when
labeled data is scarce, the recent industrial success of deep learning has revolved around supervised learning. Supervised learning is currently the focus of many recent research advances, which have shown to excel at many computer vision tasks. Top performing systems often involve very large and deep models, which are not well suited for applications with time or memory limitations. More in line with the current trends, we engage in making top performing models more efficient, by designing very deep and thin models. Since training such very deep models still appears to be a challenging task, we introduce a novel algorithm that guides the training of very thin and deep models by hinting their intermediate representations.
Very deep and thin models trained by the proposed algorithm end up extracting feature representations that are comparable or even better performing
than the ones extracted by large state-of-the-art models, while compellingly
reducing the time and memory consumption of the model.

Abstract: Egocentric images offer a hands-free way to record daily experiences and special events, where social interactions are of special interest. A natural question that arises is how to extract and track the appearance of multiple persons in a social event captured by a wearable camera. In this paper, we propose a novel method to find correspondences of multiple-faces in low temporal resolution egocentric sequences acquired through a wearable camera. This kind of sequences imposes additional challenges to the multitracking problem with respect to conventional videos. Due to the free motion of the camera and to its low temporal resolution (2 fpm), abrupt changes in the field of view, in illumination conditions and in the target location are very frequent. To overcome such a difficulty, we propose to generate, for each detected face, a set of correspondences along the whole sequence that we call tracklet and to take advantage of their redundancy to deal with both false positive face detections and unreliable tracklets. Similar tracklets are grouped into the so called extended bag-of-tracklets (eBoT), which are aimed to correspond to specific persons. Finally, a prototype tracklet is extracted for each eBoT. We validated our method over a dataset of 18.000 images from 38 egocentric sequences with 52 trackable persons and compared to the state-of-the-art methods, demonstrating its effectiveness and robustness.

Abstract: In this paper, we present a new method for egocentric video temporal segmentation based on integrating a statistical mean change detector and agglomerative clustering(AC) within an energyminimization framework. Given the tendency of most AC methods to oversegment video sequences when clustering their frames, we combine the clustering with a concept drift detection technique (ADWIN) that has rigorous guarantee of performances. ADWIN serves as a statistical upper bound for the clustering-based video segmentation. We integrate techniques in an energy-minimization framework that serves disambiguate the decision of both techniques and to complete the segmentation taking into account the temporal continuity of video frames We present experiments over egocentric sets of more than 13.000 images acquired with different wearable cameras, showing that our method outperforms state-of-the-art clustering methods.

Abstract: This paper presents the results of the UPC-UB-STP team in the 2015 MediaEval Retrieving Diverse Images Task. The goal of the challenge is to provide a ranked list of Flickr photos for a predefined set of queries. Our approach firstly generates a ranking of images based on a query-independent estimation of its relevance. Only top results are kept and iteratively re-ranked based on their intra-similarity to introduce diversity.

Abstract: This paper studies the level of complementary information conveyed by extremely high resolution LiDAR and optical images. We pursue this goal following an indirect approach via unsupervised spatial-spectral feature extraction. We used a recently presented unsupervised convolutional neural network trained to enforce both population and lifetime spar-sity in the feature representation. We derived independent and joint feature representations, and analyzed the sparsity scores and the discriminative power. Interestingly, the obtained results revealed that the RGB+LiDAR representation is no longer sparse, and the derived basis functions merge color and elevation yielding a set of more expressive colored edge filters. The joint feature representation is also more discriminative when used for clustering and topological data visualization.

Abstract: In this article we provide an overview of color name applications in computer vision. Color names are linguistic labels which humans use to communicate color. Computational color naming learns a mapping from pixels values to color names. In recent years color names have been applied to a wide variety of computer vision applications, including image classification, object recognition, texture classification, visual tracking and action recognition. Here we provide an overview of these results which show that in general color names outperform photometric invariants as a color representation.

Abstract: Describing persons and their actions is a challenging problem due to variations in pose, scale and viewpoint in real-world images. Recently, semantic pyramids approach [1] for pose normalization has shown to provide excellent results for gender and action recognition. The performance of semantic pyramids approach relies on robust image description and is therefore limited due to the use of shallow local features. In the context of object recognition [2] and object detection [3], convolutional neural networks (CNNs) or deep features have shown to improve the performance over the conventional shallow features.
We propose deep semantic pyramids for human attributes and action recognition. The method works by constructing spatial pyramids based on CNNs of different part locations. These pyramids are then combined to obtain a single semantic representation. We validate our approach on the Berkeley and 27 Human Attributes datasets for attributes classification. For action recognition, we perform experiments on two challenging datasets: Willow and PASCAL VOC 2010. The proposed deep semantic pyramids provide a significant gain of 17.2%, 13.9%, 24.3% and 22.6% compared to the standard shallow semantic pyramids on Berkeley, 27 Human Attributes, Willow and PASCAL VOC 2010 datasets respectively. Our results also show that deep semantic pyramids outperform conventional CNNs based on the full bounding box of the person. Finally, we compare our approach with state-of-the-art methods and show a gain in performance compared to best methods in literature.