Mean-Field methods for Structured Deep-Learning in Computer Vision

In recent years, Machine Learning based Computer Vision techniques made impressive progress. These algorithms proved particularly efficient for image classification or detection of isolated objects. From a probabilistic perspective, these methods can predict marginals, over single or multiple variables, independently, with high accuracy.

However, in many tasks of practical interest, we need to predict jointly several correlated variables. Practical applications include people detection in crowded scenes, image segmentation, surface reconstruction, 3D pose estimation and others. A large part of the research effort in today's computer-vision community aims at finding task-specific solutions to these problems, while leveraging the power of Deep-Learning based classifiers. In this thesis, we present our journey towards a generic and practical solution based on mean-field (MF) inference.

Mean-field is a Statistical Physics-inspired method which has long been used in Computer-Vision as a variational approximation to posterior distributions over complex Conditional Random Fields. Standard mean-field optimization is based on coordinate descent and in many situations can be impractical. We therefore propose a novel proximal gradient-based approach to optimizing the variational objective. It is naturally parallelizable and easy to implement. We prove its convergence, and then demonstrate that, in practice, it yields faster convergence and often finds better optima than more traditional mean-field optimization techniques.

Then, we show that we can replace the fully factorized distribution of mean-field by a weighted mixture of such distributions, that similarly minimizes the KL-Divergence to the true posterior. Our extension of the clamping method proposed in previous works allows us to both produce a more descriptive approximation of the true posterior and, inspired by the diverse MAP paradigms, fit a mixture of mean-field approximations. We demonstrate that this positively impacts real-world algorithms that initially relied on mean-fields.

One of the important properties of the mean-field inference algorithms is that the closed-form updates are fully differentiable operations. This naturally allows to do parameter learning by simply unrolling multiple iterations of the updates, the so-called back-mean-field algorithm. We derive a novel and efficient structured learning method for multi-modal posterior distribution based on the Multi-Modal Mean-Field approximation, which can be seamlessly combined to modern gradient-based learning methods such as CNNs.

Finally, we explore in more details the specific problem of structured learning and prediction for multiple-people detection in crowded scenes. We then present a mean-field based structured deep-learning detection algorithm that provides state of the art results on this dataset.

Object Classification and Detection in High Dimensional Feature Space

Charles Dubout

Object classification and detection aim at recognizing and localizing objects in real-world images. They are fundamental computer vision problems and a prerequisite for full scene understanding. Their difficulty lies in the large number of possible object positions and the appearance variations of object classes. This thesis improves upon several classical machine learning algorithms, enabling large computational gains in high dimensional feature space. A common trend in machine learning and computer vision research is to go large scale. In particular, the advent of huge datasets mined from the Internet, and the combination of multiple feature sources have considerably broadened the applications of computer vision. Tasks which were thought impossible a few years ago, such as human action recognition or pose estimation, automatic outdoor navigation, etc., now seem within reach. This dissertation is divided into two parts. The first one deals with the efficient training of a classifier or detector based on a large number of feature extractors, outside the control of the learning algorithm, and therefore of unknown suitability to the task at hand. More precisely, this part presents two kinds of strategies to accelerate the training of Boosting algorithms in such a context: (a) a method to better deal with the increasingly common case where features come from multiple sources (e.g. color, shape, texture, etc., in the case of images) and therefore can be partitioned into meaningful subsets; (b) new algorithms which balance at every Boosting iteration the number of weak learners and the number of training examples to look at in order to maximize the expected loss reduction. Experiments in image classification and object recognition on four standard computer vision datasets show that the adaptive techniques we propose outperform both basic sampling and state-of-the-art bandit methods. The second part deals with linear object detectors, currently the most popular class of detection systems, encompassing template matching, deformable part models, poselets, convolutional neural networks (which internally use linear filters), etc. The main bottleneck of many of those systems is the computational cost of the convolutions between the multiple rescalings of the image to process and the linear filters. We make use of properties of the Fourier transform and clever implementation strategies to obtain a speedup factor proportional to the filter size, both while training and at test time. We also introduce a few modifications to the original Deformable Part Model (DPM) of Felzenszwalb et al. improving its detection accuracy. The gains in performance are demonstrated on the well-known Pascal VOC benchmark, where an increase by one order of magnitude in the speed of said convolutions, and an average improvement of 15% in the accuracy of the detector are established.

Programme doctoral en Informatique, Communications et Information2013

Variational Methods for Human Modeling

Timur Bagautdinov

A large part of computer vision research is devoted to building models and algorithms aimed at understanding human appearance and behaviour from images and videos. Ultimately, we want to build automated systems that are at least as capable as people when it comes to interpreting humans. Most of the tasks that we want these systems to solve can be posed as a problem of inference in probabilistic models. Although probabilistic inference in general is a very hard problem of its own, there exists a very powerful class of inference algorithms, variational inference, which allows us to build efficient solutions for a wide range of problems. In this thesis, we consider a variety of computer vision problems targeted at modeling human appearance and behaviour, including detection, activity recognition, semantic segmentation and facial geometry modeling. For each of those problems, we develop novel methods that use variational inference to improve the capabilities of the existing systems. First, we introduce a novel method for detecting multiple potentially occluded people in depth images, which we call DPOM. Unlike many other approaches, our method does probabilistic reasoning jointly, and thus allows to propagate knowledge about one part of the image evidence to reason about the rest. This is particularly important in crowded scenes involving many people, since it helps to handle ambiguous situations resulting from severe occlusions. We demonstrate that our approach outperforms existing methods on multiple datasets. Second, we develop a new algorithm for variational inference that works for a large class of probabilistic models, which includes, among others, DPOM and some of the state-of-the-art models for semantic segmentation. We provide a formal proof that our method converges, and demonstrate experimentally that it brings better performance than the state-of-the-art on several real-world tasks, which include semantic segmentation and people detection. Importantly, we show that parallel variational inference in discrete random fields can be seen as a special case of proximal gradient descent, which allows us to benefit from many of the advances in gradient-based optimization. Third, we propose a unified framework for multi-human scene understanding which simultaneously solves three tasks: multi-person detection, individual action recognition and collective activity recognition. Within our framework, we introduce a novel multi-person detection scheme, which relies on variational inference and jointly refines detection hypotheses instead of relying on suboptimal post-processing. Ultimately, our model takes as an inputs a frame sequence and produces a comprehensive description of the scene. Finally, we experimentally demonstrate that our method brings better performance than the state-of-the-art. Fourth, we propose a new approach for learning facial geometry with deep probabilistic models and variational methods. Our model is based on a variational autoencoder with multiple sets of hidden variables, which are capturing various levels of deformations, ranging from global to local, high-frequency ones. We experimentally demonstrate the power of the model on a variety of fitting tasks. Our model is completely data-driven and can be learned from a relatively small number of individuals.

EPFL2018

Graph-based image representation learning

Renata Khasanova

Though deep learning (DL) algorithms are very powerful for image processing tasks, they generally require a lot of data to reach their full potential. Furthermore, there is no straightforward way to impose various properties, given by the prior knowledge about the target task, on the features extracted by a DL model. Therefore, in this thesis we propose several techniques that rely on the power of graph representations to embed prior knowledge inside the learning process. This allows to reduce the solution space and leads to faster optimization convergence and higher accuracy in the representation learning. In our first work, inspired by the ability of a human to correctly classify rotated, shifted or flipped objects, we propose an algorithm that permits to inherently encode invariance to isometric transformations of objects in an image. Our DL architecture is based on graph representations and consists of three novel layers, which we refer to as graph convolutional, dynamic pooling and statistical layers. Our experiments on the image classification tasks show that our network correctly recognizes isometrically transformed objects even though such types of transformation are not seen by the network at training time. Standard DL techniques are typically not able to succeed in solving such a problem without extensive data augmentation. Then, we propose to exploit the properties of graph-based approaches to efficiently process images with various types of projective geometry. In particular, we are interested in increasingly popular omnidirectional cameras, which have a 360 degree field of view. Despite their effectiveness, such cameras create images with specific geometric properties, which require special techniques for efficient processing. We propose an efficient way of adjusting the weights of the graph edges to adapt the filter responses to the geometric image properties introduced by omnidirectional cameras. Our experiments prove that using the proposed graph with properly adjusted edge weights permits to reach better performance as compared to using regular grid graph with equal weights. Finally, the approach described above relies on the isotropic filters, which work well within our transformation invariant architecture for image classification. However, for other problems (e.g. image compression) or even when used without dynamic pooling and statistical layers that are defined within the proposed architecture, these filters are unable to efficiently encode the information about the object. Thus, we introduce a different technique based on anisotropic filters that adapt their shape and size according to the omnidirectional image geometry. The main advantage of this approach compared to the previous one is the ability to encode the orientation of an image pattern, which is important for various tasks such as image compression. Our experiments show that our approach adapts to different image projective geometries and achieves state-of-the-art performance on image classification and compression tasks. Overall we propose several methods, which combine the power of DL and graph signal processing towards incorporating prior information about the target task inside the optimization procedure. We hope that the research efforts presented in this thesis will help the development of efficient DL algorithms that can use various types of prior knowledge to make them efficient even when the available training data is scarce.

EPFL2019