Abstract: In this paper, we propose a Text-Degradation Invariant Auto Encoder (Text-DIAE), a self-supervised model designed to tackle two tasks, text recognition (handwritten or scene-text) and document image enhancement. We start by employing a transformer-based architecture that incorporates three pretext tasks as learning objectives to be optimized during pre-training without the usage of labelled data. Each of the pretext objectives is specifically tailored for the final downstream tasks. We conduct several ablation experiments that confirm the design choice of the selected pretext tasks. Importantly, the proposed model does not exhibit limitations of previous state-of-the-art methods based on contrastive losses, while at the same time requiring substantially fewer data samples to converge. Finally, we demonstrate that our method surpasses the state-of-the-art in existing supervised and self-supervised settings in handwritten and scene text recognition and document image enhancement. Our code and trained models will be made publicly available at https://github.com/dali92002/SSL-OCR

Abstract: The curse of data labeling is a costly bottleneck in supervised deep learning, where large amounts of labeled data are needed to train intelligent systems. In onboard perception for autonomous driving, this cost corresponds to the labeling of raw data from sensors such as cameras, LiDARs, RADARs, etc. Therefore, synthetic data with automatically generated ground truth (labels) has aroused as a reliable alternative for training onboard perception models.
However, synthetic data commonly suffers from synth-to-real domain shift, i.e., models trained on the synthetic domain do not show their achievable accuracy when performing in the real world. This shift needs to be addressed by techniques falling in the realm of domain adaptation (DA).
The semi-supervised learning (SSL) paradigm can be followed to address DA. In this case, a model is trained using source data with labels (here synthetic) and leverages minimal knowledge from target data (here the real world) to generate pseudo-labels. These pseudo-labels help the training process to reduce the gap between the source and the target domains. In general, we can assume accessing both, pseudo-labels and a few amounts of human-provided labels for the target-domain data. However, the most interesting and challenging setting consists in assuming that we do not have human-provided labels at all. This setting is known as unsupervised domain adaptation (UDA). This PhD focuses on applying SSL to the UDA setting, for onboard visual tasks related to autonomous driving. We start by addressing the synth-to-real UDA problem on onboard vision-based object detection (pedestrians and cars), a critical task for autonomous driving and driving assistance. In particular, we propose to apply an SSL technique known as co-training, which we adapt to work with deep models that process a multi-modal input. The multi-modality consists of the visual appearance of the images (RGB) and their monocular depth estimation. The synthetic data we use as the source domain contains both, object bounding boxes and depth information. This prior knowledge is the
starting point for the co-training technique, which iteratively labels unlabeled real-world data and uses such pseudolabels (here bounding boxes with an assigned object class) to progressively improve the labeling results. Along this
process, two models collaborate to automatically label the images, in a way that one model compensates for the errors of the other, so avoiding error drift. While this automatic labeling process is done offline, the resulting pseudolabels can be used to train object detection models that must perform in real-time onboard a vehicle. We show that multi-modal co-training improves the labeling results compared to single-modal co-training, remaining competitive compared to human labeling.
Given the success of co-training in the context of object detection, we have also adapted this technique to a more crucial and challenging visual task, namely, onboard semantic segmentation. In fact, providing labels for a single image
can take from 30 to 90 minutes for a human labeler, depending on the content of the image. Thus, developing automatic labeling techniques for this visual task is of great interest to the automotive industry. In particular, the new co-training framework addresses synth-to-real UDA by an initial stage of self-training. Intermediate models arising from this stage are used to start the co-training procedure, for which we have elaborated an accurate collaboration policy between the two models performing the automatic labeling. Moreover, our co-training seamlessly leverages datasets from different synthetic domains. In addition, the co-training procedure is agnostic to the loss function used to train the semantic segmentation models which perform the automatic labeling. We achieve state-of-the-art results on publicly available benchmark datasets, again, remaining competitive compared to human labeling.
Finally, on the ground of our previous experience, we have designed and implemented a new SSL technique for UDA in the context of visual semantic segmentation. In this case, we mimic the labeling methodology followed by human labelers. In particular, rather than labeling full images at a time, categories of semantic classes are defined and only those are labeled in a labeling pass. In fact, different human labelers can become specialists in labeling different categories. Afterward, these per-category-labeled layers are combined to provide fully labeled images. Our technique is inspired by this methodology since we perform synth-to-real UDA per category, using the self-training stage previously developed as part of our co-training framework. The pseudo-labels obtained for each category are finally
fused to obtain fully automatically labeled images. In this context, we have also contributed to the development of a new photo-realistic synthetic dataset based on path-tracing rendering. Our new SSL technique seamlessly leverages publicly available synthetic datasets as well as this new one to obtain state-of-the-art results on synth-to-real UDA for semantic segmentation. We show that the new dataset allows us to reach better labeling accuracy than previously existing datasets, at the same time that it complements well them when combined. Moreover, we also show that the new human-inspired SSL technique outperforms co-training.

Abstract: Face Anti-spoofing (FAS) is essential to secure face recognition systems from various physical attacks. However, most of the studies lacked consideration of long-distance scenarios. Specifically, compared with FAS in traditional scenes such as phone unlocking, face payment, and self-service security inspection, FAS in long-distance such as station squares, parks, and self-service supermarkets are equally important, but it has not been sufficiently explored yet. In order to fill this gap in the FAS community, we collect a large-scale Surveillance High-Fidelity Mask (SuHiFiMask). SuHiFiMask contains 10,195 videos from 101 subjects of different age groups, which are collected by 7 mainstream surveillance cameras. Based on this dataset and protocol-3 for evaluating the robustness of the algorithm under quality changes, we organized a face presentation attack detection challenge in surveillance scenarios. It attracted 180 teams for the development phase with a total of 37 teams qualifying for the final round. The organization team re-verified and re-ran the submitted code and used the results as the final ranking. In this paper, we present an overview of the challenge, including an introduction to the dataset used, the definition of the protocol, the evaluation metrics, and the announcement of the competition results. Finally, we present the top-ranked algorithms and the research ideas provided by the competition for attack detection in long-range surveillance scenarios.

Abstract: Recent breakthroughs in single image super resolution have investigated the potential of deep Convolutional Neural Networks (CNNs) to improve performance. However, CNNs based models suffer from their limited fields and their inability to adapt to the input content. Recently, Transformer based models were presented, which demonstrated major performance gains in Natural Language Processing and Vision tasks while mitigating the drawbacks of CNNs. Nevertheless, Transformer computational complexity can increase quadratically for high-resolution images, and the fact that it ignores the original structures of the image by converting them to the 1D structure can make it problematic to capture the local context information and adapt it for real-time applications. In this paper, we present, SRFormer, an efficient yet powerful Transformer-based architecture, by making several key designs in the building of Transformer blocks and Transformer layers that allow us to consider the original structure of the image (i.e., 2D structure) while capturing both local and global dependencies without raising computational demands or memory consumption. We also present a Gated Multi-Layer Perceptron (MLP) Feature Fusion module to aggregate the features of different stages of Transformer blocks by focusing on inter-spatial relationships while adding minor computational costs to the network. We have conducted extensive experiments on several super-resolution benchmark datasets to evaluate our approach. SRFormer demonstrates superior performance compared to state-of-the-art methods from both Transformer and Convolutional networks, with an improvement margin of 0.1∼0.53dB . Furthermore, while SRFormer has almost the same model size, it outperforms SwinIR by 0.47% and inference time by half the time of SwinIR. The code will be available on GitHub.

Abstract: Recent advances in single image super-resolution (SISR) explore the power of deep convolutional neural networks (CNNs) to achieve better performance. However, most of the progress has been made by scaling CNN architectures, which usually raise computational demands and memory consumption. This makes modern architectures less applicable in practice. In addition, most CNN-based SR methods do not fully utilize the informative hierarchical features that are helpful for final image recovery. In order to address these issues, we propose a directional variance attention network (DiVANet), a computationally efficient yet accurate network for SISR. Specifically, we introduce a novel directional variance attention (DiVA) mechanism to capture long-range spatial dependencies and exploit inter-channel dependencies simultaneously for more discriminative representations. Furthermore, we propose a residual attention feature group (RAFG) for parallelizing attention and residual block computation. The output of each residual block is linearly fused at the RAFG output to provide access to the whole feature hierarchy. In parallel, DiVA extracts most relevant features from the network for improving the final output and preventing information loss along the successive operations inside the network. Experimental results demonstrate the superiority of DiVANet over the state of the art in several datasets, while maintaining relatively low computation and memory footprint. The code is available at https://github.com/pbehjatii/DiVANet.

Abstract: Visual content is being increasingly transmitted and consumed by machines rather than humans to perform automated content analysis tasks. In this paper, we propose an image preprocessor that optimizes the input image for machine consumption prior to encoding by an off-the-shelf codec designed for human consumption. To achieve a better trade-off between the accuracy of the machine analysis task and bitrate, we propose leveraging pre-extracted semantic information to improve the preprocessor’s ability to accurately identify and filter out task-irrelevant information. Furthermore, we propose a two-part loss function to optimize the preprocessor, consisted of a rate-task performance loss and a semantic distillation loss, which helps the reconstructed image obtain more information that contributes to the accuracy of the task. Experiments show that the proposed preprocessor can save up to 48.83% bitrate compared with the method without the preprocessor, and save up to 36.24% bitrate compared to existing preprocessors for machine vision.

Abstract: Developing arbitrary symbol recognition systems is a challenging endeavour. Even using content-agnostic architectures such as few-shot models, performance can be substantially improved by providing a number of well-annotated examples into training. In some contexts, transcripts of the symbols are available without any position information associated to them, which enables using line-level recognition architectures. A way of providing this position information to detection-based architectures is finding systems that can align the input symbols with the transcription. In this paper we discuss some symbol alignment techniques that are suitable for low-data scenarios and provide an insight on their perceived strengths and weaknesses. In particular, we study the usage of Connectionist Temporal Classification models, Attention-Based Sequence to Sequence models and we compare them with the results obtained on a few-shot recognition system.

Abstract: On end-to-end driving, human driving demonstrations are used to train perception-based driving models by imitation learning. This process is supervised on vehicle signals (e.g., steering angle, acceleration) but does not require extra costly supervision (human labeling of sensor data). As a representative of such vision-based end-to-end driving models, CILRS is commonly used as a baseline to compare with new driving models. So far, some latest models achieve better performance than CILRS by using expensive sensor suites and/or by using large amounts of human-labeled data for training. Given the difference in performance, one may think that it is not worth pursuing vision-based pure end-to-end driving. However, we argue that this approach still has great value and potential considering cost and maintenance. In this paper, we present CIL++, which improves on CILRS by both processing higher-resolution images using a human-inspired HFOV as an inductive bias and incorporating a proper attention mechanism. CIL++ achieves competitive performance compared to models which are more costly to develop. We propose to replace CILRS with CIL++ as a strong vision-based pure end-to-end driving baseline supervised by only vehicle signals and trained by conditional imitation learning.