Abstract: Word spotting can be defined as the pattern recognition tasked aimed at locating and retrieving a specific keyword within a document image collection without explicitly transcribing the whole corpus. Its use is particularly interesting when applied in scenarios where Optical Character Recognition performs poorly or can not be used at all. This thesis focuses on such a scenario, word spotting on historical handwritten documents that have been written by a single author or by multiple authors with a similar calligraphy.
This problem requires a visual signature that is robust to image artifacts, flexible to accommodate script variations and efficient to retrieve information in a rapid manner. For this, we have developed a set of word spotting methods that on their foundation use the well known Bag-of-Visual-Words (BoVW) representation. This representation has gained popularity among the document image analysis community to characterize handwritten words
in an unsupervised manner. However, most approaches on this field rely on a basic BoVW configuration and disregard complex encoding and spatial representations. We determine which BoVW configurations provide the best performance boost to a spotting system.
Then, we extend the segmentation-based word spotting, where word candidates are given a priori, to segmentation-free spotting. The proposed approach seeds the document images with overlapping word location candidates and characterizes them with a BoVW signature. Retrieval is achieved comparing the query and candidate signatures and returning the locations that provide a higher consensus. This is a simple but powerful approach that requires a more compact signature than in a segmentation-based scenario. We first
project the BoVW signature into a reduced semantic topics space and then compress it further using Product Quantizers. The resulting signature only requires a few dozen bytes, allowing us to index thousands of pages on a common desktop computer. The final system still yields a performance comparable to the state-of-the-art despite all the information loss during the compression phases.
Afterwards, we also study how to combine different modalities of information in order to create a query-by-X spotting system where, words are indexed using an information modality and queries are retrieved using another. We consider three different information modalities: visual, textual and audio. Our proposal is to create a latent feature space where features which are semantically related are projected onto the same topics. Creating thus a new feature space where information from different modalities can be compared. Later, we consider the codebook generation and descriptor encoding problem. The codebooks used to encode the BoVW signatures are usually created using an unsupervised clustering algorithm and, they require to test multiple parameters to determine which configuration is best for a certain document collection. We propose a semantic clustering algorithm which allows to estimate the best parameter from data. Since gather annotated data is costly, we use synthetically generated word images. The resulting codebook is database agnostic, i. e. a codebook that yields a good performance on document collections that use the same script. We also propose the use of an additional codebook to approximate descriptors and reduce the descriptor encoding
complexity to sub-linear.
Finally, we focus on the problem of signatures dimensionality. We propose a new symbol probability signature where each bin represents the probability that a certain symbol is present a certain location of the word image. This signature is extremely compact and combined with compression techniques can represent word images with just a few bytes per signature.

Abstract: Manually annotating images to develop vision models has been a major bottleneck
since computer vision and machine learning started to walk together. This has
been more evident since computer vision falls on the shoulders of data-hungry
deep learning techniques. When addressing on-board perception for autonomous
driving, the curse of data annotation is exacerbated due to the use of additional
sensors such as LiDAR. Therefore, any approach aiming at reducing such a timeconsuming and costly work is of high interest for addressing autonomous driving
and, in fact, for any application requiring some sort of artificial perception. In the
last decade, it has been shown that leveraging from synthetic data is a paradigm
worth to pursue in order to minimizing manual data annotation. The reason is
that the automatic process of generating synthetic data can also produce different
types of associated annotations (e.g. object bounding boxes for synthetic images
and LiDAR pointclouds, pixel/point-wise semantic information, etc.). Directly
using synthetic data for training deep perception models may not be the definitive
solution in all circumstances since it can appear a synth-to-real domain shift. In
this context, this work focuses on leveraging synthetic data to alleviate manual
annotation for three perception tasks related to driving assistance and autonomous
driving. In all cases, we assume the use of deep convolutional neural networks
(CNNs) to develop our perception models.
The first task addresses traffic sign recognition (TSR), a kind of multi-class
classification problem. We assume that the number of sign classes to be recognized
must be suddenly increased without having annotated samples to perform the
corresponding TSR CNN re-training. We show that leveraging synthetic samples of
such new classes and transforming them by a generative adversarial network (GAN)
trained on the known classes (i.e. without using samples from the new classes), it is
possible to re-train the TSR CNN to properly classify all the signs for a ∼ 1/4 ratio of
new/known sign classes. The second task addresses on-board 2D object detection,
focusing on vehicles and pedestrians. In this case, we assume that we receive a set
of images without the annotations required to train an object detector, i.e. without
object bounding boxes. Therefore, our goal is to self-annotate these images so
that they can later be used to train the desired object detector. In order to reach
this goal, we leverage from synthetic data and propose a semi-supervised learning
approach based on the co-training idea. In fact, we use a GAN to reduce the synthto-real domain shift before applying co-training. Our quantitative results show
that co-training and GAN-based image-to-image translation complement each
other up to allow the training of object detectors without manual annotation, and still almost reaching the upper-bound performances of the detectors trained from
human annotations. While in previous tasks we focus on vision-based perception,
the third task we address focuses on LiDAR pointclouds. Our initial goal was to
develop a 3D object detector trained on synthetic LiDAR-style pointclouds. While
for images we may expect synth/real-to-real domain shift due to differences in
their appearance (e.g. when source and target images come from different camera
sensors), we did not expect so for LiDAR pointclouds since these active sensors
factor out appearance and provide sampled shapes. However, in practice, we have
seen that it can be domain shift even among real-world LiDAR pointclouds. Factors
such as the sampling parameters of the LiDARs, the sensor suite configuration onboard the ego-vehicle, and the human annotation of 3D bounding boxes, do induce
a domain shift. We show it through comprehensive experiments with different
publicly available datasets and 3D detectors. This redirected our goal towards the
design of a GAN for pointcloud-to-pointcloud translation, a relatively unexplored
topic.
Finally, it is worth to mention that all the synthetic datasets used for these three
tasks, have been designed and generated in the context of this PhD work and will
be publicly released. Overall, we think this PhD presents several steps forward to
encourage leveraging synthetic data for developing deep perception models in the
field of driving assistance and autonomous driving.

Abstract: Developing human-like digital characters is a challenging task since humans are used to recognizing our fellows, and find the computed generated characters inadequately humanized. To fulfill the standards of the videogame and digital film productions it is necessary to model and animate these characters the most closely to human beings. However, it is an arduous and expensive task, since many artists and specialists are required to work on a single character. Therefore, to fulfill these requirements we found an interesting option to study the automatic creation of detailed characters through inexpensive setups. In this work, we develop novel techniques to bring detailed characters by combining different aspects that stand out when developing realistic characters, skin detail, facial hairs, expressions, and microexpressions. We examine each of the mentioned areas with the aim of automatically recover each of the parts without user interaction nor training data. We study the problems for their robustness but also for the simplicity of the setup, preferring single-image with uncontrolled illumination and methods that can be easily computed with the commodity of a standard laptop. A detailed face with wrinkles and skin details is vital to develop a realistic character. In this work, we introduce our method to automatically describe facial wrinkles on the image and transfer to the recovered base face. Then we advance to facial hair recovery by resolving a fitting problem with a novel parametrization model. As of last, we develop a mapping function that allows transfer expressions and microexpressions between different meshes, which provides realistic animations to our detailed mesh. We cover all the mentioned points with the focus on key aspects as (i) how to describe skin wrinkles in a simple and straightforward manner, (ii) how to recover 3D from 2D detections, (iii) how to recover and model facial hair from 2D to 3D, (iv) how to transfer expressions between models holding both skin detail and facial hair, (v) how to perform all the described actions without training data nor user interaction. In this work, we present our proposals to solve these aspects with an efficient and simple setup. We validate our work with several datasets both synthetic and real data, prooving remarkable results even in challenging cases as occlusions as glasses, thick beards, and indeed working with different face topologies like single-eyed cyclops.

Abstract: We introduce the problem of weakly supervised MultiObject Tracking and Segmentation, i.e. joint weakly supervised instance segmentation and multi-object tracking, in which we do not provide any kind of mask annotation.
To address it, we design a novel synergistic training strategy by taking advantage of multi-task learning, i.e. classification and tracking tasks guide the training of the unsupervised instance segmentation. For that purpose, we extract weak foreground localization information, provided by
Grad-CAM heatmaps, to generate a partial ground truth to learn from. Additionally, RGB image level information is employed to refine the mask prediction at the edges of the
objects. We evaluate our method on KITTI MOTS, the most representative benchmark for this task, reducing the performance gap on the MOTSP metric between the fully supervised and weakly supervised approach to just 12% and 12.7 % for cars and pedestrians, respectively.

Abstract: In class-incremental learning, an agent with limited resources needs to learn a sequence of classification tasks, forming an ever growing classification problem, with the constraint of not being able to access data from previous tasks. The main difference with task-incremental learning, where a task-ID is available at inference time, is that the learner also needs to perform crosstask discrimination, i.e. distinguish between classes that have not been seen together. Approaches to tackle this problem are numerous and mostly make use of an external memory (buffer) of non-negligible size. In this paper, we ablate the learning of crosstask features and study its influence on the performance of basic replay strategies used for class-IL. We also define a new forgetting measure for class-incremental learning, and see that forgetting is not the principal cause of low performance. Our experimental results show that future algorithms for class-incremental learning should not only prevent forgetting, but also aim to improve the quality of the cross-task features. This is especially important when the number of classes per task is small.

Abstract: Images and videos are pervasive in our life and communication. With advances in smart and portable devices, high capacity communication networks and high definition cinema, image and video compression are more relevant than ever. Traditional block-based linear transform codecs such as JPEG, H.264/AVC or the recent H.266/VVC are carefully designed to meet not only the rate-distortion criteria, but also the practical requirements of applications.
Recently, a new paradigm based on deep neural networks (i.e., neural image/video compression) has become increasingly popular due to its ability to learn powerful nonlinear transforms and other coding tools directly from data instead of being crafted by humans, as was usual in previous coding formats. While achieving excellent rate-distortion performance, these approaches are still limited mostly to research environments due to heavy models and other practical limitations, such as being limited to function on a particular rate and due to high memory and computational cost. In this thesis, we study these practical limitations, and designing more practical neural image compression approaches.
After analyzing the differences between traditional and neural image compression, our first contribution is the modulated autoencoder (MAE), a framework that includes a mechanism to provide multiple rate-distortion options within a single model with comparable performance to independent models. In a second contribution, we propose the slimmable compressive autoencoder (SlimCAE), which in addition to variable rate, can optimize the complexity of the model and thus reduce significantly the memory and computational burden.
Modern generative models can learn custom image transformation directly from suitable datasets following encoder-decoder architectures, task known as image-to-image (I2I) translation. Building on our previous work, we study the problem of distributed I2I translation, where the latent representation is transmitted through a binary channel and decoded in a remote receiving side. We also propose a variant that can perform both translation and the usual autoencoding functionality.
Finally, we also consider neural video compression, where the autoencoder is typically augmented with temporal prediction via motion compensation. One of the main bottlenecks of that framework is the optical flow module that estimates the displacement to predict the next frame. Focusing on this module, we propose a method that improves the accuracy of the optical flow estimation and a simplified variant that reduces the computational cost.
Key words: neural image compression, neural video compression, optical flow, practical neural image compression, compressive autoencoders, image-to-image translation, deep learning.

Abstract: From the early stages of Computer Vision, scene reconstruction has been one of the most studied topics leading to a wide variety of new discoveries and applications. Object grasping and manipulation, localization and mapping, or even visual effect generation are different examples of applications in which scene reconstruction has taken an important role for industries such as robotics, factory automation, or audio visual production. However, scene reconstruction is an extensive topic that can be approached in many different ways with already existing solutions that effectively work in controlled environments. Formally, the problem of scene reconstruction can be formulated as a sequence of independent processes which compose a pipeline. In this thesis, we analyse some parts of the reconstruction pipeline from which we contribute with novel methods using Convolutional Neural Networks (CNN) proposing innovative solutions that consider the optimisation of the methods in an end-to-end fashion. First, we review the state of the art of classical local features detectors and descriptors and contribute with two novel methods that inherently improve pre-existing solutions in the scene reconstruction pipeline.
It is a fact that computer science and software engineering are two fields that usually go hand in hand and evolve according to mutual needs making easier the design of complex and efficient algorithms. For this reason, we contribute with Kornia, a library specifically designed to work with classical computer vision techniques along with deep neural networks. In essence, we created a framework that eases the design of complex pipelines for computer vision algorithms so that can be included within neural networks and be used to backpropagate gradients throw a common optimisation framework. Finally, in the last chapter of this thesis we develop the aforementioned concept of designing end-to-end systems with classical projective geometry. Thus, we contribute with a solution to the problem of synthetic view generation by hallucinating novel views from high deformable cloths objects using a geometry aware end-to-end system. To summarize, in this thesis we demonstrate that with a proper design that combine classical geometric computer vision methods with deep learning techniques can lead to improve pre-existing solutions for the problem of scene reconstruction.