Deep Learning approaches have brought solutions, with impressive performance, to general classification problems where wealthy of annotated data are provided for training. In contrast, less progress has been made in continual learning of a set of non-stationary classes, mainly when applied to unsupervised problems with streaming data. Here, we propose a novel incremental learning approach which combines a deep features encoder with an Open-Set Dynamic Ensembles of SVM, to tackle the problem of identifying individuals of interest (IoI) from streaming face data. From a simple weak classifier trained on a few video-frames, our method can use unsupervised operational data to enhance recognition. Our approach adapts to new patterns avoiding catastrophic forgetting and partially heals itself from miss-adaptation. Besides, to better comply with real world conditions, the system was designed to operate in an open-set setting. Results show a benefit of up to 15% F1-score increase respect to non-adaptive state-of-the-art methods.

The absence of a previous collaborative manual enrolment represents a significant handicap towards designing a face verification system for face re-identification purposes. In this scenario, the system must learn the target identity incrementally, using data from the video stream during the operational authentication phase. So, manual labelling cannot be assumed apart from the first few frames. On the other hand, even the most advanced methods trained on large-scale and unconstrained datasets suffer performance degradation when no adaptation to specific contexts is performed. This work proposes an adaptive face verification system, for the continuous re-identification of target identity, within the framework of incremental unsupervised learning. Our Dynamic Ensemble of SVM is capable of incorporating non-labelled information to improve the performance of any model, even when its initial performance is modest. The proposal uses the self-training approach and is compared against other classification techniques within this same approach. Results show promising behaviour in terms of both knowledge acquisition and impostor robustness.

Video-to-video face recognition in unconstrained conditions is still a very challenging problem, as the combination of several factors leads to an in general low-quality of facial frames. Besides, in some real contexts, the availability of labelled samples is limited, or data is streaming or it is only available temporarily due to storage constraints or privacy issues. In these cases, dealing with learning as an unsupervised incremental process is a feasible option. This work proposes a system based on dynamic ensembles of SVM's, which uses the ideas of self-training to perform adaptive Video-to-video face identification. The only label requirements of the system are a few frames (5 in our experiments) directly taken from the video-surveillance stream. The system will autonomously use additional video-frames to update and improve the initial model in an unsupervised way. Results show a significant improvement in comparison to other state-of-the-art static models.

A benchmark of saliency models performance with a synthetic image dataset is provided. Model performance is evaluated through saliency metrics as well as the influence of model inspiration and consistency with human psychophysics. SID4VAM is composed of 230 synthetic images, with known salient regions. Images were generated with 15 distinct types of low-level features (e.g. orientation, brightness, color, size...) with a target-distractor pop-out type of synthetic patterns. We have used Free-Viewing and Visual Search task instructions and 7 feature contrasts for each feature category. Our study reveals that state-of-the-art Deep Learning saliency models do not perform well with synthetic pattern images, instead, models with Spectral/Fourier inspiration outperform others in saliency metrics and are more consistent with human psychophysical experimentation. This study proposes a new way to evaluate saliency models in the forthcoming literature, accounting for synthetic images with uniquely low-level feature contexts, distinct from previous eye tracking image datasets.

General dynamic scenes involve multiple rigid and flexible objects, with relative and common motion, camera induced or not. The complexity of the motion events together with their strong spatio-temporal correlations make the estimation of dynamic visual saliency a big computational challenge. In this work, we propose a computational model of saliency based on the assumption that perceptual relevant information is carried by high-order statistical structures. Through whitening, we completely remove the second-order information (correlations and variances) of the data, gaining access to the relevant information. The proposed approach is an analytically tractable and computationally simple framework which we call Dynamic Adaptive Whitening Saliency (AWS-D). For model assessment, the provided saliency maps were used to predict the fixations of human observers over six public video datasets, and also to reproduce the human behavior under certain psychophysical experiments (dynamic pop-out). The results demonstrate that AWS-D beats state-of-the-art dynamic saliency models, and suggest that the model might contain the basis to understand the key mechanisms of visual saliency. Experimental evaluation was performed using an extension to video of the well-known methodology for static images, together with a bootstrap permutation test (random label hypothesis) which yields additional information about temporal evolution of the metrics statistical significance.

We propose a new statistic deformable model that we call discriminant snake for 3D reconstruction in volumetric images. Our discriminant snake generalises the classical snake attracted by edge points; it deforms due to a generalised contour representation. The snake selects and classifies image features by a parametric classifier and each snaxel deforms to minimise the dissimilarity between the learned and found image features inside the feature space. We apply our statistic snake to segment anatomical organs and the results are very encouraging.

An alternative to classical image segmentation based on active contour techniques is discussed. The approach is based on deformable contours which evolve until reaching a final desired location. The contour shift is guided by external information from the image under consideration which attracts them towards the target characteristics (intensity, extremes, edges,...) and by internal forces which try to maintain the smoothness of the contour curve. These forces act on each pixel of the contours, resulting in a high degree of freedom for the contour evolution and provide a high flexibility for the evolution dynamics of the snakes which allows the solution of complex tasks as is the case for topologic transformations. This, along with the use of only local information will allow the algorithm implementation into an array of processors leading towards "pixel-level" contour processing.