Abstract: Visual recognition is one of the main problems in computer vision that attempts to solve image understanding by deciding what objects are in images. This problem can be computationally solved by using relevant sets of visual features, such as edges, corners, color or more complex object parts. This thesis contributes to how color features have to be represented for recognition tasks.
Image features can be extracted following two different approaches. A first approach is defining handcrafted descriptors of images which is then followed by a learning scheme to classify the content (named flat schemes in Kruger et al. (2013). In this approach, perceptual considerations are habitually used to define efficient color features. Here we propose a new flat color descriptor based on the extension of color channels to boost the representation of spatio-chromatic contrast that surpasses state-of-the-art approaches. However, flat schemes present a lack of generality far away from the capabilities of biological systems. A second approach proposes evolving these flat schemes into a hierarchical process, like in the visual cortex. This includes an automatic process to learn optimal features. These deep schemes, and more specifically Convolutional Neural Networks (CNNs), have shown an impressive performance to solve various vision problems. However, there is a lack of understanding about the internal representation obtained, as a result of automatic learning. In this thesis we propose a new methodology to explore the internal representation of trained CNNs by defining the Neuron Feature as a visualization of the intrinsic features encoded in each individual neuron. Additionally, and inspired by physiological techniques, we propose to compute different neuron selectivity indexes (e.g., color, class, orientation or symmetry, amongst others) to label and classify the full CNN neuron population to understand learned representations.
Finally, using the proposed methodology, we show an in-depth study on how color is represented on a specific CNN, trained for object recognition, that competes with primate representational abilities (Cadieu et al (2014)). We found several parallelisms with biological visual systems: (a) a significant number of color selectivity neurons throughout all the layers; (b) an opponent and low frequency representation of color oriented edges and a higher sampling of frequency selectivity in brightness than in color in 1st layer like in V1; (c) a higher sampling of color hue in the second layer aligned to observed hue maps in V2; (d) a strong color and shape entanglement in all layers from basic features in shallower layers (V1 and V2) to object and background shapes in deeper layers (V4 and IT); and (e) a strong correlation between neuron color selectivities and color dataset bias.

Abstract: Developing a text detection method which is invariant to scripts in natural scene images is a challeng- ing task due to different geometrical structures of various scripts. Besides, multi-oriented of text lines in natural scene images make the problem more challenging. This paper proposes to explore ring radius transform (RRT) for text detection in multi-oriented and multi-script environments. The method finds component regions based on convex hull to generate radius matrices using RRT. It is a fact that RRT pro- vides low radius values for the pixels that are near to edges, constant radius values for the pixels that represent stroke width, and high radius values that represent holes created in background and convex hull because of the regular structures of text components. We apply k -means clustering on the radius matrices to group such spatially coherent regions into individual clusters. Then the proposed method studies the radius values of such cluster components that are close to the centroid and far from the cen- troid to detect text components. Furthermore, we have developed a Bangla dataset (named as ISI-UM dataset) and propose a semi-automatic system for generating its ground truth for text detection of arbi- trary orientations, which can be used by the researchers for text detection and recognition in the future. The ground truth will be released to public. Experimental results on our ISI-UM data and other standard datasets, namely, ICDAR 2013 scene, SVT and MSRA data, show that the proposed method outperforms the existing methods in terms of multi-lingual and multi-oriented text detection ability.

Abstract: In this paper, we propose an automatic method for constructing an active shape model (ASM) to segment the complete cardiac left ventricle in 3D ultrasound (3DUS) images, which avoids costly manual landmarking. The automatic construction of the ASM has already been addressed in the literature; however, the direct application of these methods to 3DUS is hampered by a high level of noise and artefacts. Therefore, we propose to construct the ASM by fusing the multidetector computed tomography data, to learn the shape, with the artificially generated 3DUS, in order to learn the neighbourhood of the boundaries. Our artificial images were generated by two approaches: a faster one that does not take into account the geometry of the transducer, and a more comprehensive one, implemented in Field II toolbox. The segmentation accuracy of our ASM was evaluated on 20 patients with left-ventricular asynchrony, demonstrating plausibility of the approach.

Abstract: We present an application of gesture recognition using an extension of Dynamic Time Warping (DTW) to recognize behavioural patterns of Attention Deficit Hyperactivity Disorder (ADHD). We propose an extension of DTW using one-class classifiers in order to be able to encode the variability of a gesture category, and thus, perform an alignment between a gesture sample and a gesture class. We model the set of gesture samples of a certain gesture category using either GMMs or an approximation of Convex Hulls. Thus, we add a theoretical contribution to classical warping path in DTW by including local modeling of intra-class gesture variability. This methodology is applied in a clinical context, detecting a group of ADHD behavioural patterns defined by experts in psychology/psychiatry, to provide support to clinicians in the diagnose procedure. The proposed methodology is tested on a novel multi-modal dataset (RGB plus Depth) of ADHD children recordings with behavioural patterns. We obtain satisfying results when compared to standard state-of-the-art approaches in the DTW context.

Abstract: A key topic in classification is the accuracy loss produced when the data distribution in the training (source) domain differs from that in the testing (target) domain. This is being recognized as a very relevant problem for many
computer vision tasks such as image classification, object detection, and object category recognition. In this paper, we present a novel domain adaptation method that leverages multiple target domains (or sub-domains) in a hierarchical adaptation tree. The core idea is to exploit the commonalities and differences of the jointly considered target domains.
Given the relevance of structural SVM (SSVM) classifiers, we apply our idea to the adaptive SSVM (A-SSVM), which only requires the target domain samples together with the existing source-domain classifier for performing the desired adaptation. Altogether, we term our proposal as hierarchical A-SSVM (HA-SSVM).
As proof of concept we use HA-SSVM for pedestrian detection, object category recognition and face recognition. In the former we apply HA-SSVM to the deformable partbased model (DPM) while in the rest HA-SSVM is applied to multi-category classifiers. We will show how HA-SSVM is effective in increasing the detection/recognition accuracy with respect to adaptation strategies that ignore the structure of the target data. Since, the sub-domains of the target data are not always known a priori, we shown how HA-SSVM can incorporate sub-domain discovery for object category recognition.

Abstract: Quantitative assessment of calcified atherosclerotic volume within the coronary artery wall is vital for cardiac interventional procedures. The goal of this study is to automatically measure the calcium volume, given the borders of coronary vessel wall for all the frames of the intravascular ultrasound (IVUS) video. Three soft computing fuzzy classification techniques were adapted namely Fuzzy c-Means (FCM), K-means, and Hidden Markov Random Field (HMRF) for automated segmentation of calcium regions and volume computation. These methods were benchmarked against previously developed threshold-based method. IVUS image data sets (around 30,600 IVUS frames) from 15 patients were collected using 40 MHz IVUS catheter (Atlantis® SR Pro, Boston Scientific®, pullback speed of 0.5 mm/s). Calcium mean volume for FCM, K-means, HMRF and threshold-based method were 37.84 ± 17.38 mm3, 27.79 ± 10.94 mm3, 46.44 ± 19.13 mm3 and 35.92 ± 16.44 mm3 respectively. Cross-correlation, Jaccard Index and Dice Similarity were highest between FCM and threshold-based method: 0.99, 0.92 ± 0.02 and 0.95 + 0.02 respectively. Student’s t-test, z-test and Wilcoxon-test are also performed to demonstrate consistency, reliability and accuracy of the results. Given the vessel wall region, the system reliably and automatically measures the calcium volume in IVUS videos. Further, we validated our system against a trained expert using scoring: K-means showed the best performance with an accuracy of 92.80 %. Out procedure and protocol is along the line with method previously published clinically.

Abstract: Facial expressions are an important way through which humans interact socially. Building a system capable of automatically recognizing facial expressions from images and video has been an intense field of study in recent years. Interpreting such expressions remains challenging and much research is needed about the way they relate to human affect. This paper presents a general overview of automatic RGB, 3D, thermal and multimodal facial expression analysis. We define a new taxonomy for the field, encompassing all steps from face detection to facial expression recognition, and describe and classify the state of the art methods accordingly. We also present the important datasets and the bench-marking of most influential methods. We conclude with a general discussion about trends, important questions and future lines of research.

Abstract: During face-to-face human interaction, nonverbal communication plays a fundamental role. A relevant aspect that takes part during social interactions is represented by mirroring, in which a person tends to mimic the non-verbal behavior (head and body gestures, vocal prosody, etc.) of the counterpart. In this paper, we introduce a computer vision-based system to detect mirroring in dyadic social interactions with the use of a wearable platform. In our context, mirroring is inferred as simultaneous head noddings displayed by the interlocutors. Our approach consists of the following steps: (1) facial features extraction; (2) facial features stabilization; (3) head nodding recognition; and (4) mirroring detection. Our system achieves a mirroring detection accuracy of 72% on a custom mirroring dataset.