Lluis Gomez, Ali Furkan Biten, Ruben Tito, Andres Mafla, Marçal Rusiñol, Ernest Valveny, et al. (2021). Multimodal grid features and cell pointers for scene text visual question answering. PRL - Pattern Recognition Letters, 150, 242–249.
Abstract: This paper presents a new model for the task of scene text visual question answering. In this task questions about a given image can only be answered by reading and understanding scene text. Current state of the art models for this task make use of a dual attention mechanism in which one attention module attends to visual features while the other attends to textual features. A possible issue with this is that it makes difficult for the model to reason jointly about both modalities. To fix this problem we propose a new model that is based on an single attention mechanism that attends to multi-modal features conditioned to the question. The output weights of this attention module over a grid of multi-modal spatial features are interpreted as the probability that a certain spatial location of the image contains the answer text to the given question. Our experiments demonstrate competitive performance in two standard datasets with a model that is faster than previous methods at inference time. Furthermore, we also provide a novel analysis of the ST-VQA dataset based on a human performance study. Supplementary material, code, and data is made available through this link.
|
Minesh Mathew, Lluis Gomez, Dimosthenis Karatzas, & C.V. Jawahar. (2021). Asking questions on handwritten document collections. IJDAR - International Journal on Document Analysis and Recognition, 24, 235–249.
Abstract: This work addresses the problem of Question Answering (QA) on handwritten document collections. Unlike typical QA and Visual Question Answering (VQA) formulations where the answer is a short text, we aim to locate a document snippet where the answer lies. The proposed approach works without recognizing the text in the documents. We argue that the recognition-free approach is suitable for handwritten documents and historical collections where robust text recognition is often difficult. At the same time, for human users, document image snippets containing answers act as a valid alternative to textual answers. The proposed approach uses an off-the-shelf deep embedding network which can project both textual words and word images into a common sub-space. This embedding bridges the textual and visual domains and helps us retrieve document snippets that potentially answer a question. We evaluate results of the proposed approach on two new datasets: (i) HW-SQuAD: a synthetic, handwritten document image counterpart of SQuAD1.0 dataset and (ii) BenthamQA: a smaller set of QA pairs defined on documents from the popular Bentham manuscripts collection. We also present a thorough analysis of the proposed recognition-free approach compared to a recognition-based approach which uses text recognized from the images using an OCR. Datasets presented in this work are available to download at docvqa.org.
|
Vacit Oguz Yazici. (2022). Towards Smart Fashion: Visual Recognition of Products and Attributes (Joost Van de Weijer, & Arnau Ramisa, Eds.). Ph.D. thesis, IMPRIMA, .
Abstract: Artificial intelligence is innovating the fashion industry by proposing new applications and solutions to the problems encountered by researchers and engineers working in the industry. In this thesis, we address three of these problems. In the first part of the thesis, we tackle the problem of multi-label image classification which is very related to fashion attribute recognition. In the second part of the thesis, we address two problems that are specific to fashion. Firstly, we address the problem of main product detection which is the task of associating correct image parts (e.g. bounding boxes) with the fashion product being sold. Secondly, we address the problem of color naming for multicolored fashion items. The task of multi-label image classification consists in assigning various concepts such as objects or attributes to images. Usually, there are dependencies that can be learned between the concepts to capture label correlations (chair and table classes are more likely to co-exist than chair and giraffe).
If we treat the multi-label image classification problem as an orderless set prediction problem, we can exploit recurrent neural networks (RNN) to capture label correlations. However, RNNs are trained to predict ordered sequences of tokens, so if the order of the predicted sequence is different than the order of the ground truth sequence, there will be penalization although the predictions are correct. Therefore, in the first part of the thesis, we propose an orderless loss function which will order the labels in the ground truth sequence dynamically in a way that the minimum loss is achieved. This results in a significant improvement of RNN models on multi-label image classification over the previous methods.
However, RNNs suffer from long term dependencies when the cardinality of set grows bigger. The decoding process might stop early if the current hidden state cannot find any object and outputs the termination token. This would cause the remaining classes not to be predicted and lower recall metric. Transformers can be used to avoid the long term dependency problem exploiting their selfattention modules that process sequential data simultaneously. Consequently, we propose a novel transformer model for multi-label image classification which surpasses the state-of-the-art results by a large margin.
In the second part of thesis, we focus on two fashion-specific problems. Main product detection is the task of associating image parts with the fashion product that is being sold, generally using associated textual metadata (product title or description). Normally, in fashion e-commerces, products are represented by multiple images where a person wears the product along with other fashion items. If all the fashion items in the images are marked with bounding boxes, we can use the textual metadata to decide which item is the main product. The initial work treated each of these images independently, discarding the fact that they all belong to the same product. In this thesis, we represent the bounding boxes from all the images as nodes in a fully connected graph. This allows the algorithm to learn relations between the nodes during training and take the entire context into account for the final decision. Our algorithm results in a significant improvement of the state-ofthe-art.
Moreover, we address the problem of color naming for multicolored fashion items, which is a challenging task due to the external factors such as illumination changes or objects that act as clutter. In the context of multi-label classification, the vaguely defined lines between the classes in the color space cause ambiguity. For example, a shade of blue which is very close to green might cause the model to incorrectly predict the color blue and green at the same time. Based on this, models trained for color naming are expected to recognize the colors and their quantities in both single colored and multicolored fashion items. Therefore, in this thesis, we propose a novel architecture with an additional head that explicitly estimates the number of colors in fashion items. This removes the ambiguity problem and results in better color naming performance.
|
Manisha Das, Deep Gupta, Petia Radeva, & Ashwini M. Bakde. (2021). Multi-scale decomposition-based CT-MR neurological image fusion using optimized bio-inspired spiking neural model with meta-heuristic optimization. IMA - International Journal of Imaging Systems and Technology, 31(4), 2170–2188.
Abstract: Multi-modal medical image fusion plays an important role in clinical diagnosis and works as an assistance model for clinicians. In this paper, a computed tomography-magnetic resonance (CT-MR) image fusion model is proposed using an optimized bio-inspired spiking feedforward neural network in different decomposition domains. First, source images are decomposed into base (low-frequency) and detail (high-frequency) layer components. Low-frequency subbands are fused using texture energy measures to capture the local energy, contrast, and small edges in the fused image. High-frequency coefficients are fused using firing maps obtained by pixel-activated neural model with the optimized parameters using three different optimization techniques such as differential evolution, cuckoo search, and gray wolf optimization, individually. In the optimization model, a fitness function is computed based on the edge index of resultant fused images, which helps to extract and preserve sharp edges available in the source CT and MR images. To validate the fusion performance, a detailed comparative analysis is presented among the proposed and state-of-the-art methods in terms of quantitative and qualitative measures along with computational complexity. Experimental results show that the proposed method produces a significantly better visual quality of fused images meanwhile outperforms the existing methods.
|
Alina Matei, Andreea Glavan, Petia Radeva, & Estefania Talavera. (2021). Towards Eating Habits Discovery in Egocentric Photo-Streams. ACCESS - IEEE Access, 9, 17495–17506.
Abstract: Eating habits are learned throughout the early stages of our lives. However, it is not easy to be aware of how our food-related routine affects our healthy living. In this work, we address the unsupervised discovery of nutritional habits from egocentric photo-streams. We build a food-related behavioral pattern discovery model, which discloses nutritional routines from the activities performed throughout the days. To do so, we rely on Dynamic-Time-Warping for the evaluation of similarity among the collected days. Within this framework, we present a simple, but robust and fast novel classification pipeline that outperforms the state-of-the-art on food-related image classification with a weighted accuracy and F-score of 70% and 63%, respectively. Later, we identify days composed of nutritional activities that do not describe the habits of the person as anomalies in the daily life of the user with the Isolation Forest method. Furthermore, we show an application for the identification of food-related scenes when the camera wearer eats in isolation. Results have shown the good performance of the proposed model and its relevance to visualize the nutritional habits of individuals.
|
Dorota Kaminska, Kadir Aktas, Davit Rizhinashvili, Danila Kuklyanov, Abdallah Hussein Sham, Sergio Escalera, et al. (2021). Two-stage Recognition and Beyond for Compound Facial Emotion Recognition. ELEC - Electronics, 10(22), 2847.
Abstract: Facial emotion recognition is an inherently complex problem due to individual diversity in facial features and racial and cultural differences. Moreover, facial expressions typically reflect the mixture of people’s emotional statuses, which can be expressed using compound emotions. Compound facial emotion recognition makes the problem even more difficult because the discrimination between dominant and complementary emotions is usually weak. We have created a database that includes 31,250 facial images with different emotions of 115 subjects whose gender distribution is almost uniform to address compound emotion recognition. In addition, we have organized a competition based on the proposed dataset, held at FG workshop 2020. This paper analyzes the winner’s approach—a two-stage recognition method (1st stage, coarse recognition; 2nd stage, fine recognition), which enhances the classification of symmetrical emotion labels.
Keywords: compound emotion recognition; facial expression recognition; dominant and complementary emotion recognition; deep learning
|
Hugo Bertiche, Meysam Madadi, & Sergio Escalera. (2021). PBNS: Physically Based Neural Simulation for Unsupervised Garment Pose Space Deformation. ACM Transactions on Graphics, 40(6), 1–14.
Abstract: We present a methodology to automatically obtain Pose Space Deformation (PSD) basis for rigged garments through deep learning. Classical approaches rely on Physically Based Simulations (PBS) to animate clothes. These are general solutions that, given a sufficiently fine-grained discretization of space and time, can achieve highly realistic results. However, they are computationally expensive and any scene modification prompts the need of re-simulation. Linear Blend Skinning (LBS) with PSD offers a lightweight alternative to PBS, though, it needs huge volumes of data to learn proper PSD. We propose using deep learning, formulated as an implicit PBS, to unsupervisedly learn realistic cloth Pose Space Deformations in a constrained scenario: dressed humans. Furthermore, we show it is possible to train these models in an amount of time comparable to a PBS of a few sequences. To the best of our knowledge, we are the first to propose a neural simulator for cloth.
While deep-based approaches in the domain are becoming a trend, these are data-hungry models. Moreover, authors often propose complex formulations to better learn wrinkles from PBS data. Supervised learning leads to physically inconsistent predictions that require collision solving to be used. Also, dependency on PBS data limits the scalability of these solutions, while their formulation hinders its applicability and compatibility. By proposing an unsupervised methodology to learn PSD for LBS models (3D animation standard), we overcome both of these drawbacks. Results obtained show cloth-consistency in the animated garments and meaningful pose-dependant folds and wrinkles. Our solution is extremely efficient, handles multiple layers of cloth, allows unsupervised outfit resizing and can be easily applied to any custom 3D avatar.
|
Clementine Decamps, Alexis Arnaud, Florent Petitprez, Mira Ayadi, Aurelia Baures, Lucile Armenoult, et al. (2021). DECONbench: a benchmarking platform dedicated to deconvolution methods for tumor heterogeneity quantification. BMC Bioinformatics, 22, 473.
Abstract: Quantification of tumor heterogeneity is essential to better understand cancer progression and to adapt therapeutic treatments to patient specificities. Bioinformatic tools to assess the different cell populations from single-omic datasets as bulk transcriptome or methylome samples have been recently developed, including reference-based and reference-free methods. Improved methods using multi-omic datasets are yet to be developed in the future and the community would need systematic tools to perform a comparative evaluation of these algorithms on controlled data.
|
Javier Marin, & Sergio Escalera. (2021). SSSGAN: Satellite Style and Structure Generative Adversarial Networks. Remote Sensing, 13(19), 3984.
Abstract: This work presents Satellite Style and Structure Generative Adversarial Network (SSGAN), a generative model of high resolution satellite imagery to support image segmentation. Based on spatially adaptive denormalization modules (SPADE) that modulate the activations with respect to segmentation map structure, in addition to global descriptor vectors that capture the semantic information in a vector with respect to Open Street Maps (OSM) classes, this model is able to produce
consistent aerial imagery. By decoupling the generation of aerial images into a structure map and a carefully defined style vector, we were able to improve the realism and geodiversity of the synthesis with respect to the state-of-the-art baseline. Therefore, the proposed model allows us to control the generation not only with respect to the desired structure, but also with respect to a geographic area.
|
Meysam Madadi, Sergio Escalera, Xavier Baro, & Jordi Gonzalez. (2022). End-to-end Global to Local CNN Learning for Hand Pose Recovery in Depth data. IETCV - IET Computer Vision, 16(1), 50–66.
Abstract: Despite recent advances in 3D pose estimation of human hands, especially thanks to the advent of CNNs and depth cameras, this task is still far from being solved. This is mainly due to the highly non-linear dynamics of fingers, which make hand model training a challenging task. In this paper, we exploit a novel hierarchical tree-like structured CNN, in which branches are trained to become specialized in predefined subsets of hand joints, called local poses. We further fuse local pose features, extracted from hierarchical CNN branches, to learn higher order dependencies among joints in the final pose by end-to-end training. Lastly, the loss function used is also defined to incorporate appearance and physical constraints about doable hand motion and deformation. Finally, we introduce a non-rigid data augmentation approach to increase the amount of training depth data. Experimental results suggest that feeding a tree-shaped CNN, specialized in local poses, into a fusion network for modeling joints correlations and dependencies, helps to increase the precision of final estimations, outperforming state-of-the-art results on NYU and SyntheticHand datasets.
Keywords: Computer vision; data acquisition; human computer interaction; learning (artificial intelligence); pose estimation
|
Victor M. Campello, Polyxeni Gkontra, Cristian Izquierdo, Carlos Martin-Isla, Alireza Sojoudi, Peter M. Full, et al. (2021). Multi-Centre, Multi-Vendor and Multi-Disease Cardiac Segmentation: The M&Ms Challenge. TMI - IEEE Transactions on Medical Imaging, 40(12), 3543–3554.
Abstract: The emergence of deep learning has considerably advanced the state-of-the-art in cardiac magnetic resonance (CMR) segmentation. Many techniques have been proposed over the last few years, bringing the accuracy of automated segmentation close to human performance. However, these models have been all too often trained and validated using cardiac imaging samples from single clinical centres or homogeneous imaging protocols. This has prevented the development and validation of models that are generalizable across different clinical centres, imaging conditions or scanner vendors. To promote further research and scientific benchmarking in the field of generalizable deep learning for cardiac segmentation, this paper presents the results of the Multi-Centre, Multi-Vendor and Multi-Disease Cardiac Segmentation (M&Ms) Challenge, which was recently organized as part of the MICCAI 2020 Conference. A total of 14 teams submitted different solutions to the problem, combining various baseline models, data augmentation strategies, and domain adaptation techniques. The obtained results indicate the importance of intensity-driven data augmentation, as well as the need for further research to improve generalizability towards unseen scanner vendors or new imaging protocols. Furthermore, we present a new resource of 375 heterogeneous CMR datasets acquired by using four different scanner vendors in six hospitals and three different countries (Spain, Canada and Germany), which we provide as open-access for the community to enable future research in the field.
|
Meysam Madadi, Hugo Bertiche, & Sergio Escalera. (2021). Deep unsupervised 3D human body reconstruction from a sparse set of landmarks. IJCV - International Journal of Computer Vision, 129, 2499–2512.
Abstract: In this paper we propose the first deep unsupervised approach in human body reconstruction to estimate body surface from a sparse set of landmarks, so called DeepMurf. We apply a denoising autoencoder to estimate missing landmarks. Then we apply an attention model to estimate body joints from landmarks. Finally, a cascading network is applied to regress parameters of a statistical generative model that reconstructs body. Our set of proposed loss functions allows us to train the network in an unsupervised way. Results on four public datasets show that our approach accurately reconstructs the human body from real world mocap data.
|
Meysam Madadi, Hugo Bertiche, Wafa Bouzouita, Isabelle Guyon, & Sergio Escalera. (2021). Learning Cloth Dynamics: 3D+Texture Garment Reconstruction Benchmark. In Proceedings of Machine Learning Research (Vol. 133, pp. 57–76).
Abstract: Human avatars are important targets in many computer applications. Accurately tracking, capturing, reconstructing and animating the human body, face and garments in 3D are critical for human-computer interaction, gaming, special effects and virtual reality. In the past, this has required extensive manual animation. Regardless of the advances in human body and face reconstruction, still modeling, learning and analyzing human dynamics need further attention. In this paper we plan to push the research in this direction, e.g. understanding human dynamics in 2D and 3D, with special attention to garments. We provide a large-scale dataset (more than 2M frames) of animated garments with variable topology and type, calledCLOTH3D++. The dataset contains RGBA video sequences paired with its corresponding 3D data. We pay special care to garment dynamics and realistic rendering of RGB data, including lighting, fabric type and texture. With this dataset, we hold a competition at NeurIPS2020. We design three tracks so participants can compete to develop the best method to perform 3D garment reconstruction in a sequence from (1) 3D-to-3D garments, (2) RGB-to-3D garments, and (3) RGB-to-3D garments plus texture. We also provide a baseline method, based on graph convolutional networks, for each track. Baseline results show that there is a lot of room for improvements. However, due to the challenging nature of the problem, no participant could outperform the baselines.
|
Swathikiran Sudhakaran, Sergio Escalera, & Oswald Lanz. (2021). Learning to Recognize Actions on Objects in Egocentric Video with Attention Dictionaries. TPAMI - IEEE Transactions on Pattern Analysis and Machine Intelligence, .
Abstract: We present EgoACO, a deep neural architecture for video action recognition that learns to pool action-context-object descriptors from frame level features by leveraging the verb-noun structure of action labels in egocentric video datasets. The core component of EgoACO is class activation pooling (CAP), a differentiable pooling operation that combines ideas from bilinear pooling for fine-grained recognition and from feature learning for discriminative localization. CAP uses self-attention with a dictionary of learnable weights to pool from the most relevant feature regions. Through CAP, EgoACO learns to decode object and scene context descriptors from video frame features. For temporal modeling in EgoACO, we design a recurrent version of class activation pooling termed Long Short-Term Attention (LSTA). LSTA extends convolutional gated LSTM with built-in spatial attention and a re-designed output gate. Action, object and context descriptors are fused by a multi-head prediction that accounts for the inter-dependencies between noun-verb-action structured labels in egocentric video datasets. EgoACO features built-in visual explanations, helping learning and interpretation. Results on the two largest egocentric action recognition datasets currently available, EPIC-KITCHENS and EGTEA, show that by explicitly decoding action-context-object descriptors, EgoACO achieves state-of-the-art recognition performance.
|
Fatemeh Noroozi, Ciprian Corneanu, Dorota Kamińska, Tomasz Sapiński, Sergio Escalera, & Gholamreza Anbarjafari. (2021). Survey on Emotional Body Gesture Recognition. TAC - IEEE Transactions on Affective Computing, 12(2), 505–523.
Abstract: Automatic emotion recognition has become a trending research topic in the past decade. While works based on facial expressions or speech abound, recognizing affect from body gestures remains a less explored topic. We present a new comprehensive survey hoping to boost research in the field. We first introduce emotional body gestures as a component of what is commonly known as “body language” and comment general aspects as gender differences and culture dependence. We then define a complete framework for automatic emotional body gesture recognition. We introduce person detection and comment static and dynamic body pose estimation methods both in RGB and 3D. We then comment the recent literature related to representation learning and emotion recognition from images of emotionally expressive gestures. We also discuss multi-modal approaches that combine speech or face with body gestures for improved emotion recognition. While pre-processing methodologies (e.g. human detection and pose estimation) are nowadays mature technologies fully developed for robust large scale analysis, we show that for emotion recognition the quantity of labelled data is scarce, there is no agreement on clearly defined output spaces and the representations are shallow and largely based on naive geometrical representations.
|