Learning and Exploiting Camera Geometry for Computer Vision

This work explores the use of statistical methods in describing and estimating camera poses, as well as the information feedback loop between camera pose and object detection. Surging development in robotics and computer vision has pushed the need for algorithms that infer, understand, and utilize information about the position and orientation of the sensor platforms when observing and/or interacting with their environment.

The first contribution of this thesis is the development of a set of statistical tools for representing and estimating the uncertainty in object poses. A distribution for representing the joint uncertainty over multiple object positions and orientations is described, called the mirrored normal-Bingham distribution. This distribution generalizes both the normal distribution in Euclidean space, and the Bingham distribution on the unit hypersphere. It is shown to inherit many of the convenient properties of these special cases: it is the maximum-entropy distribution with fixed second moment, and there is a generalized Laplace approximation whose result is the mirrored normal-Bingham distribution. This distribution and approximation method are demonstrated by deriving the analytical approximation to the wrapped-normal distribution. Further, it is shown how these tools can be used to represent the uncertainty in the result of a bundle adjustment problem.

Another application of these methods is illustrated as part of a novel camera pose estimation algorithm based on object detections. The autocalibration task is formulated as a bundle adjustment problem using prior distributions over the 3D points to enforce the objects' structure and their relationship with the scene geometry. This framework is very flexible and enables the use of off-the-shelf computational tools to solve specialized autocalibration problems. Its performance is evaluated using a pedestrian detector to provide head and foot location observations, and it proves much faster and potentially more accurate than existing methods.

Finally, the information feedback loop between object detection and camera pose estimation is closed by utilizing camera pose information to improve object detection in scenarios with significant perspective warping. Methods are presented that allow the inverse perspective mapping traditionally applied to images to be applied instead to features computed from those images. For the special case of HOG-like features, which are used by many modern object detection systems, these methods are shown to provide substantial performance benefits over unadapted detectors while achieving real-time frame rates, orders of magnitude faster than comparable image warping methods.

The statistical tools and algorithms presented here are especially promising for mobile cameras, providing the ability to autocalibrate and adapt to the camera pose in real time. In addition, these methods have wide-ranging potential applications in diverse areas of computer vision, robotics, and imaging.

Continual learning for computer vision applications

One of the most visionary goals of Artificial Intelligence is to create a system able to mimic and eventually surpass the intelligence observed in biological systems including, ambitiously, the one observed in humans. The main distinctive strength of humans is their ability to build a deep understanding of the world by learning continuously and drawing from their experiences. This ability, which is found in various degrees in all intelligent biological beings, allows them to adapt and properly react to changes by incrementally expanding and refining their knowledge. Arguably, achieving this ability is one of the main goals of Artificial Intelligence and a cornerstone towards the creation of intelligent artificial agents. Modern Deep Learning approaches allowed researchers and industries to achieve great advancements towards the resolution of many long-standing problems in areas like Computer Vision and Natural Language Processing. However, while this current age of renewed interest in AI allowed for the creation of extremely useful applications, a concerningly limited effort is being directed towards the design of systems able to learn continuously. The biggest problem that hinders an AI system from learning incrementally is the catastrophic forgetting phenomenon. This phenomenon, which was discovered in the 90s, naturally occurs in Deep Learning architectures where classic learning paradigms are applied when learning incrementally from a stream of experiences. This dissertation revolves around the Continual Learning field, a sub-field of Machine Learning research that has recently made a comeback following the renewed interest in Deep Learning approaches. This work will focus on a comprehensive view of continual learning by considering algorithmic, benchmarking, and applicative aspects of this field. This dissertation will also touch on community aspects such as the design and creation of research tools aimed at supporting Continual Learning research, and the theoretical and practical aspects concerning public competitions in this field.

Computer vision in aseptic pharmaceutical manufacturing: a deep learning-based approach to vial inspection and pack formation during a lyophilizer loading cycle

Vision systems are powerful tools playing an increasingly important role in modern industry, to detect errors and maintain product standards. With the enlarged availability of affordable industrial cameras, computer vision algorithms have been increasingly applied in industrial manufacturing processes monitoring. Until a few years ago, industrial computer vision applications relied only on ad-hoc algorithms designed for the specific object and acquisition setup being monitored, with a strong focus on co-designing the acquisition and processing pipeline. Deep learning has overcome these limits providing greater flexibility and faster re-configuration. In this work, the process to be inspected consists in vials’ pack formation entering a freeze-dryer, which is a common scenario in pharmaceutical active ingredient packaging lines. To ensure that the machine produces proper packs, a vision system is installed at the entrance of the freeze-dryer to detect eventual anomalies with execution times compatible with the production specifications. Other constraints come from sterility and safety standards required in pharmaceutical manufacturing. This work presents an overview about the production line, with particular focus on the vision system designed, and about all trials conducted to obtain the final performance. Transfer learning, alleviating the requirement for a large number of training data, combined with data augmentation methods, consisting in the generation of synthetic images, were used to effectively increase the performances while reducing the cost of data acquisition and annotation. The proposed vision algorithm is composed by two main subtasks, designed respectively to vials counting and discrepancy detection. The first one was trained on more than 23k vials (about 300 images) and tested on 5k more (about 75 images), whereas 60 training images and 52 testing images were used for the second one.

A Methodology for GUI Layer Redefinition Through Virtualization and Computer Vision

Nowadays despite improvements in usability and intuitiveness users have to adapt to the proposed systems to satisfy their needs. For instance, they must learn how to achieve tasks, how to interact with the system, and fulfill system's specifications. This paper proposes an approach to improve this situation enabling graphical user interface redefinition through virtualization and computer vision with the aim of increasing the system's usability. To achieve this goal the approach is based on enriched task models, virtualization and picture-driven computing.

Vision based obstacle detection for all-terrain robots

Dissertação apresentada na Faculdade de Ciências e Tecnologia da Universidade Nova de Lisboa para obtenção do grau de Mestre em Engenharia Electrotécnica e de Computadores

Vision based trail detection for all-terrain robots

Dissertação apresentada na Faculdade de Ciências e Tecnologia da Universidade Nova de Lisboa para obtenção do grau de Mestre em Engenharia Electrotécnica e de Computadores

Detection of matchings in a sequence of underwater images through texture analysis

This paper presents an approach to ameliorate the reliability of the correspondence points relating two consecutive images of a sequence. The images are especially difficult to handle, since they have been acquired by a camera looking at the sea floor while carried by an underwater robot. Underwater images are usually difficult to process due to light absorption, changing image radiance and lack of well-defined features. A new approach based on gray-level region matching and selective texture analysis significantly improves the matching reliability

Reconeixement d’escenes de la ciutat de Girona a partir del contingut d’imatges

Desenvolupament una aplicació informàtica basada en un sistema de visió per computador, la qual permeti donar una resposta en forma d'informació a partir d'una query d'una imatge que conté una escena o objecte en concret de manera que permeti reconèixer els objectes que apareixen en una imatge per llavors donar informació referent al contingut de la imatge a l’usuari que ha fet la consulta. Resumint, es tracta d’analitzar, dissenyar i construir un sistemade visió per computador capaç de reconèixer objectes d’interès en imatges

Classificació i reconeixement de vídeo

El principal objectiu d’aquest projecte és aconseguir classificar diferents vídeos d’esports segons la seva categoria. Els cercadors de text creen un vocabulari segons el significat de les diferents paraules per tal de poder identificar un document. En aquest projecte es va fer el mateix però mitjançant paraules visuals. Per exemple, es van intentar englobar com a una única paraula les diferents rodes que apareixien en els cotxes de rally. A partir de la freqüència amb què apareixien les paraules dels diferents grups dins d’una imatge vàrem crear histogrames de vocabulari que ens permetien tenir una descripció de la imatge. Per classificar un vídeo es van utilitzar els histogrames que descrivien els seus fotogrames. Com que cada histograma es podia considerar un vector de valors enters vàrem optar per utilitzar una màquina classificadora de vectors: una Support vector machine o SVM

Correcció postural esportiva utilitzant un sensor Kinect

Aquest projecte s’emmarca dins de l’àmbit de la visió per computador, concretament en la utilització de dades de profunditat obtingudes a través d’un emissor i sensor de llum infraroja.El propòsit principal d’aquest projecte és mostrar com adaptar aquestes tecnologies, a l’abast de qualsevol particular, de forma que un usuari durant la pràctica d’una activitat esportiva concreta, rebi informació visual continua dels moviments i gestos incorrectes que està realitzant, en base a uns paràmetres prèviament establerts.L’objectiu d’aquest projecte consisteix en fer una lectura constant en temps real d’una persona practicant una selecció de diverses activitats esportives estàtiques utilitzant un sensor Kinect. A través de les dades obtingudes pel sensor Kinect i utilitzant les llibreries de “skeleton traking” proporcionades per Microsoft s’haurà d’interpretar les dades posturals obtingudes per cada tipus d’esport i indicar visualment i d’una manera intuïtiva els errors que està cometent en temps real, de manera que es vegi clarament a quina part del seu cos realitza un moviment incorrecte per tal de poder corregir-lo ràpidament. El entorn de desenvolupament que s’utilitza per desenvolupar aquesta aplicació es Microsoft Viusal Studio 2010.El llenguatge amb el qual es treballarà sobre Microsoft Visual Studio 2010 és C#

Performance evaluation of 3D computer vision techniques

This work presents the implementation and comparison of three different techniques of three-dimensional computer vision as follows: • Stereo vision - correlation between two 2D images • Sensorial fusion - use of different sensors: camera 2D + ultrasound sensor (1D); • Structured light The computer vision techniques herein presented took into consideration the following characteristics: • Computational effort ( elapsed time for obtain the 3D information); • Influence of environmental conditions (noise due to a non uniform lighting, overlighting and shades); • The cost of the infrastructure for each technique; • Analysis of uncertainties, precision and accuracy. The option of using the Matlab software, version 5.1, for algorithm implementation of the three techniques was due to the simplicity of their commands, programming and debugging. Besides, this software is well known and used by the academic community, allowing the results of this work to be obtained and verified. Examples of three-dimensional vision applied to robotic assembling tasks ("pick-and-place") are presented.

Computer vision techniques in seafood quality control

The application of computer vision based quality control has been slowly but steadily gaining importance mainly due to its speed in achieving results and also greatly due to its non- destnictive nature of testing. Besides, in food applications it also does not contribute to contamination. However, computer vision applications in quality control needs the application of an appropriate software for image analysis. Eventhough computer vision based quality control has several advantages, its application has limitations as to the type of work to be done, particularly so in the food industries. Selective applications, however, can be highly advantageous and very accurate.Computer vision based image analysis could be used in morphometric measurements of fish with the same accuracy as the existing conventional method. The method is non-destructive and non-contaminating thus providing anadvantage in seafood processing.The images could be stored in archives and retrieved at anytime to carry out morphometric studies for biologists.Computer vision and subsequent image analysis could be used in measurements of various food products to assess uniformity of size. One product namely cutlet and product ingredients namely coating materials such as bread crumbs and rava were selected for the study. Computer vision based image analysis was used in the measurements of length, width and area of cutlets. Also the width of coating materials like bread crumbs was measured.Computer imaging and subsequent image analysis can be very effectively used in quality evaluations of product ingredients in food processing. Measurement of width of coating materials could establish uniformity of particles or the lack of it. The application of image analysis in bacteriological work was also done

Detection of matchings in a sequence of underwater images through texture analysis

This paper presents an approach to ameliorate the reliability of the correspondence points relating two consecutive images of a sequence. The images are especially difficult to handle, since they have been acquired by a camera looking at the sea floor while carried by an underwater robot. Underwater images are usually difficult to process due to light absorption, changing image radiance and lack of well-defined features. A new approach based on gray-level region matching and selective texture analysis significantly improves the matching reliability

Reconeixement d’escenes de la ciutat de Girona a partir del contingut d’imatges

Desenvolupament una aplicació informàtica basada en un sistema de visió per computador, la qual permeti donar una resposta en forma d'informació a partir d'una query d'una imatge que conté una escena o objecte en concret de manera que permeti reconèixer els objectes que apareixen en una imatge per llavors donar informació referent al contingut de la imatge a l’usuari que ha fet la consulta. Resumint, es tracta d’analitzar, dissenyar i construir un sistema de visió per computador capaç de reconèixer objectes d’interès en imatges

Classificació i reconeixement de vídeo

El principal objectiu d’aquest projecte és aconseguir classificar diferents vídeos d’esports segons la seva categoria. Els cercadors de text creen un vocabulari segons el significat de les diferents paraules per tal de poder identificar un document. En aquest projecte es va fer el mateix però mitjançant paraules visuals. Per exemple, es van intentar englobar com a una única paraula les diferents rodes que apareixien en els cotxes de rally. A partir de la freqüència amb què apareixien les paraules dels diferents grups dins d’una imatge vàrem crear histogrames de vocabulari que ens permetien tenir una descripció de la imatge. Per classificar un vídeo es van utilitzar els histogrames que descrivien els seus fotogrames. Com que cada histograma es podia considerar un vector de valors enters vàrem optar per utilitzar una màquina classificadora de vectors: una Support vector machine o SVM