Abstract: At the beginning of the 21th century, traffic accidents have become a major problem not only for developed countries but also for emerging ones. As in other scientific areas in which Artificial Intelligence is becoming a key actor, advanced driver assistance systems, and concretely pedestrian protection systems based on Computer Vision, are becoming a strong topic of research aimed at improving the safety of pedestrians. However, the challenge is of considerable complexity due to the varying appearance of humans (e.g., clothes, size, aspect ratio, shape, etc.), the dynamic nature of on-board systems and the unstructured moving environments that urban scenarios represent. In addition, the required performance is demanding both in terms of computational time and detection rates. In this thesis, instead of focusing on improving specific tasks as it is frequent in the literature, we present a global approach to the problem. Such a global overview starts by the proposal of a generic architecture to be used as a framework both to review the literature and to organize the studied techniques along the thesis. We then focus the research on tasks such as foreground segmentation, object classification and refinement following a general viewpoint and exploring aspects that are not usually analyzed. In order to perform the experiments, we also present a novel pedestrian dataset that consists of three subsets, each one addressed to the evaluation of a different specific task in the system. The results presented in this thesis not only end with a proposal of a pedestrian detection system but also go one step beyond by pointing out new insights, formalizing existing and proposed algorithms, introducing new techniques and evaluating their performance, which we hope will provide new foundations for future research in the area.

Abstract: Humansmove their eyes in order to learn visual representations of the world. These eye movements depend on distinct factors, either by the scene that we perceive or by our own decisions. To select what is relevant to attend is part of our survival mechanisms and the way we build reality, as we constantly react both consciously and unconsciously to all the stimuli that is projected into our eyes. In this thesis we try to explain (1) how we move our eyes, (2) how to build machines that understand visual information and deploy eyemovements, and (3) how to make these machines understand tasks in order to decide for eye movements.
(1) We provided the analysis of eye movement behavior elicited by low-level feature distinctiveness with a dataset of 230 synthetically-generated image patterns. A total of 15 types of stimuli has been generated (e.g. orientation, brightness, color, size, etc.), with 7 feature contrasts for each feature category. Eye-tracking data was collected from 34 participants during the viewing of the dataset, using Free-Viewing and Visual Search task instructions. Results showed that saliency is predominantly and distinctively influenced by: 1. feature type, 2. feature contrast, 3. Temporality of fixations, 4. task difficulty and 5. center bias. From such dataset (SID4VAM), we have computed a benchmark of saliency models by testing performance using psychophysical patterns. Model performance has been evaluated considering model inspiration and consistency with human psychophysics. Our study reveals that state-of-the-art Deep Learning saliency models do not performwell with synthetic pattern images, instead, modelswith Spectral/Fourier inspiration outperform others in saliency metrics and are more consistent with human psychophysical experimentation.
(2) Computations in the primary visual cortex (area V1 or striate cortex) have long been hypothesized to be responsible, among several visual processing mechanisms, of bottom-up visual attention (also named saliency). In order to validate this hypothesis, images from eye tracking datasets have been processed with a biologically plausible model of V1 (named Neurodynamic SaliencyWaveletModel or NSWAM). Following Li’s neurodynamic model, we define V1’s lateral connections with a network of firing rate neurons, sensitive to visual features such as brightness, color, orientation and scale. Early subcortical processes (i.e. retinal and thalamic) are functionally simulated. The resulting saliency maps are generated from the model output, representing the neuronal activity of V1 projections towards brain areas involved in eye movement control. We want to pinpoint that our unified computational architecture is able to reproduce several visual processes (i.e. brightness, chromatic induction and visual discomfort) without applying any type of training or optimization and keeping the same parametrization. The model has been extended (NSWAM-CM) with an implementation of the cortical magnification function to define the retinotopical projections towards V1, processing neuronal activity for each distinct view during scene observation. Novel computational definitions of top-down inhibition (in terms of inhibition of return and selection mechanisms), are also proposed to predict attention in Free-Viewing and Visual Search conditions. Results show that our model outperforms other biologically-inpired models of saliency prediction as well as to predict visual saccade sequences, specifically for nature and synthetic images. We also show how temporal and spatial characteristics of inhibition of return can improve prediction of saccades, as well as how distinct search strategies (in terms of feature-selective or category-specific inhibition) predict attention at distinct image contexts.
(3) Although previous scanpath models have been able to efficiently predict saccades during Free-Viewing, it is well known that stimulus and task instructions can strongly affect eye movement patterns. In particular, task priming has been shown to be crucial to the deployment of eye movements, involving interactions between brain areas related to goal-directed behavior, working and long-termmemory in combination with stimulus-driven eyemovement neuronal correlates. In our latest study we proposed an extension of the Selective Tuning Attentive Reference Fixation ControllerModel based on task demands (STAR-FCT), describing novel computational definitions of Long-TermMemory, Visual Task Executive and Task Working Memory. With these modules we are able to use textual instructions in order to guide the model to attend to specific categories of objects and/or places in the scene. We have designed our memorymodel by processing a visual hierarchy of low- and high-level features. The relationship between the executive task instructions and the memory representations has been specified using a tree of semantic similarities between the learned features and the object category labels. Results reveal that by using this model, the resulting object localizationmaps and predicted saccades have a higher probability to fall inside the salient regions depending on the distinct task instructions compared to saliency.

Abstract: Deep learning has made tremendous progress in the last decade due to the explosion of training data and computational power. Through end-to-end training on a
large dataset, image representations are more discriminative than the previously
used hand-crafted features. However, for many real-world applications, training
and testing on a single dataset is not realistic, as the test distribution may change over time. Continuous learning takes this situation into account, where the learner must adapt to a sequence of tasks, each with a different distribution. If you would naively continue training the model with a new task, the performance of the model would drop dramatically for the previously learned data. This phenomenon is known as catastrophic forgetting.
Many approaches have been proposed to address this problem, which can be divided into three main categories: regularization-based approaches, rehearsal-based
approaches, and parameter isolation-based approaches. However, most of the existing works focus on image classification tasks and many other computer vision tasks
have not been well-explored in the continual learning setting. Therefore, in this
thesis, we study continual learning for image generation, object re-identification,
and object counting.
For the image generation problem, since the model can generate images from the previously learned task, it is free to apply rehearsal without any limitation. We developed two methods based on generative replay. The first one uses the generated image for joint training together with the new data. The second one is based on
output pixel-wise alignment. We extensively evaluate these methods on several
benchmarks.
Next, we study continual learning for object Re-Identification (ReID). Although
most state-of-the-art methods of ReID and continual ReID use softmax-triplet loss,
we found that it is better to solve the ReID problem from a meta-learning perspective because continual learning of reID can benefit a lot from the generalization of metalearning. We also propose a distillation loss and found that the removal of the positive pairs before the distillation loss is critical.
Finally, we study continual learning for the counting problem. We study the mainstream method based on density maps and propose a new approach for density
map distillation. We found that fixing the counter head is crucial for the continual learning of object counting. To further improve results, we propose an adaptor to adapt the changing feature extractor for the fixed counter head. Extensive evaluation shows that this results in improved continual learning performance.

Abstract: For humans, the recognition of objects is an almost instantaneous, precise and
extremely adaptable process. Furthermore, we have the innate capability to learn
new object classes from only few examples. The human brain lowers the complexity
of the incoming data by filtering out part of the information and only processing
those things that capture our attention. This, mixed with our biological predisposition to respond to certain shapes or colors, allows us to recognize in a simple
glance the most important or salient regions from an image. This mechanism can
be observed by analyzing on which parts of images subjects place attention; where
they fix their eyes when an image is shown to them. The most accurate way to
record this behavior is to track eye movements while displaying images.
Computational saliency estimation aims to identify to what extent regions or
objects stand out with respect to their surroundings to human observers. Saliency
maps can be used in a wide range of applications including object detection, image
and video compression, and visual tracking. The majority of research in the field has
focused on automatically estimating saliency maps given an input image. Instead, in
this thesis, we set out to incorporate saliency maps in an object recognition pipeline:
we want to investigate whether saliency maps can improve object recognition
results.
In this thesis, we identify several problems related to visual saliency estimation.
First, to what extent the estimation of saliency can be exploited to improve the
training of an object recognition model when scarce training data is available. To
solve this problem, we design an image classification network that incorporates
saliency information as input. This network processes the saliency map through a
dedicated network branch and uses the resulting characteristics to modulate the
standard bottom-up visual characteristics of the original image input. We will refer to this technique as saliency-modulated image classification (SMIC). In extensive
experiments on standard benchmark datasets for fine-grained object recognition,
we show that our proposed architecture can significantly improve performance,
especially on dataset with scarce training data.
Next, we address the main drawback of the above pipeline: SMIC requires an
explicit saliency algorithm that must be trained on a saliency dataset. To solve this,
we implement a hallucination mechanism that allows us to incorporate the saliency
estimation branch in an end-to-end trained neural network architecture that only
needs the RGB image as an input. A side-effect of this architecture is the estimation
of saliency maps. In experiments, we show that this architecture can obtain similar
results on object recognition as SMIC but without the requirement of ground truth
saliency maps to train the system.
Finally, we evaluated the accuracy of the saliency maps that occur as a sideeffect of object recognition. For this purpose, we use a set of benchmark datasets
for saliency evaluation based on eye-tracking experiments. Surprisingly, the estimated saliency maps are very similar to the maps that are computed from human
eye-tracking experiments. Our results show that these saliency maps can obtain
competitive results on benchmark saliency maps. On one synthetic saliency dataset
this method even obtains the state-of-the-art without the need of ever having seen
an actual saliency image for training.

Abstract: The increasing ubiquitousness of digital information in our daily lives has positioned
video as a favored information vehicle, and given rise to an astonishing generation of
social media and surveillance footage. This raises a series of technological demands
for automatic video understanding and management, which together with the compromising attentional limitations of human operators, have motivated the research
community to guide its steps towards a better attainment of such capabilities. As
a result, current trends on cognitive vision promise to recognize complex events and
self-adapt to different environments, while managing and integrating several types of
knowledge. Future directions suggest to reinforce the multi-modal fusion of information sources and the communication with end-users.
In this thesis we tackle the problem of recognizing and describing meaningful
events in video sequences from different domains, and communicating the resulting
knowledge to end-users by means of advanced interfaces for human–computer interaction. This problem is addressed by designing the high-level modules of a cognitive
vision framework exploiting ontological knowledge. Ontologies allow us to define the
relevant concepts in a domain and the relationships among them; we prove that the
use of ontologies to organize, centralize, link, and reuse different types of knowledge
is a key factor in the materialization of our objectives.
The proposed framework contributes to: (i) automatically learn the characteristics
of different scenarios in a domain; (ii) reason about uncertain, incomplete, or vague
information from visual –camera’s– or linguistic –end-user’s– inputs; (iii) derive plausible interpretations of complex events from basic spatiotemporal developments; (iv)
facilitate natural interfaces that adapt to the needs of end-users, and allow them to
communicate efficiently with the system at different levels of interaction; and finally,
(v) find mechanisms to guide modeling processes, maintain and extend the resulting
models, and to exploit multimodal resources synergically to enhance the former tasks.
We describe a holistic methodology to achieve these goals. First, the use of prior
taxonomical knowledge is proved useful to guide MAP-MRF inference processes in
the automatic identification of semantic regions, with independence of a particular scenario. Towards the recognition of complex video events, we combine fuzzy
metric-temporal reasoning with SGTs, thus assessing high-level interpretations from
spatiotemporal data. Here, ontological resources like T–Boxes, onomasticons, or factual databases become useful to derive video indexing and retrieval capabilities, and
also to forward highlighted content to smart user interfaces. There, we explore the
application of ontologies to discourse analysis and cognitive linguistic principles, or scene augmentation techniques towards advanced communication by means of natural language dialogs and synthetic visualizations. Ontologies become fundamental to
coordinate, adapt, and reuse the different modules in the system.
The suitability of our ontological framework is demonstrated by a series of applications that especially benefit the field of smart video surveillance, viz. automatic generation of linguistic reports about the content of video sequences in multiple natural
languages; content-based filtering and summarization of these reports; dialogue-based
interfaces to query and browse video contents; automatic learning of semantic regions
in a scenario; and tools to evaluate the performance of components and models in the
system, via simulation and augmented reality.

Abstract: Automatic understanding of human activity and action is very important and challenging research area of Computer Vision with wide applications in video surveillance, motion analysis, virtual reality interfaces, video indexing, content based video retrieval, HCI and health care. This thesis presents a series of techniques to solve the problem of human action recognition in video. First approach towards this goal is based on a probabilistic optimization model of body parts using Hidden Markov Model. This strong model based approach is able to distinguish between similar actions by only considering the body parts having major contributions to the actions. In next approach, we apply a weak model based human detector and actions are represented by Bag-of-key poses model to capture the human pose changes during the actions. To tackle the problem of human action recognition in complex scenes, a selective spatio-temporal interest point (STIP) detector is proposed by using a mechanism similar to that of the non-classical receptive field inhibition that is exhibited by most oriented selective neuron in the primary visual cortex. An extension of the selective STIP detector is applied to multi-view action recognition system by introducing a novel 4D STIPs (3D space + time). Finally, we use our STIP detector on large scale continuous visual event recognition problem and propose a novel generalized max-margin Hough transformation framework for activity detection