While vision transformers have achieved impressive results, effectively and efficiently accelerating these models can further boost performances. In this work, we propose a dense/sparse training framework to obtain a unified model, enabling weight sharing across various token densities. Thus one model offers a range of accuracy and throughput tradeoffs for different applications. Besides, we introduce adaptive token pruning to optimize the patch token sparsity based on the input image. In addition, we investigate knowledge distillation to enhance token selection capability in early transformer modules. Sparse adaptive image Transformer (SaiT) offers varying levels of model acceleration by merely changing the token sparsity on the fly. Specifically, SaiT reduces the computation complexity (FLOPs) by 39% - 43% and increases the throughput by 67% - 91% with less than 0.5% accuracy loss for various vision transformer models. Meanwhile, the same model also provides the zero accuracy drop option by skipping the sparsification step. SaiT achieves better accuracy and computation tradeoffs than state-of-the-art transformer and convolutional models.
Though image transformers have shown competitive results with convolutional neural networks in computer vision tasks, lacking inductive biases such as locality still poses problems in terms of model efficiency especially for embedded applications. In this work, we address this issue by introducing attention masks to incorporate spatial locality into self-attention heads. Local dependencies are captured efficiently with masked attention heads along with global dependencies captured by unmasked attention heads. With Masked attention image Transformer - MaiT, top-1 accuracy increases by up to 1.7% compared to CaiT with fewer parameters and FLOPs, and the throughput improves by up to 1.5X compared to Swin. Encoding locality with attention masks is model agnostic, and thus it applies to monolithic, hierarchical, or other novel transformer architectures.
Automatic assessment and understanding of facial skin condition have several applications, including the early detection of underlying health problems, lifestyle and dietary treatment, skin-care product recommendation, etc. Selfies in the wild serve as an excellent data resource to democratize skin quality assessment, but suffer from several data collection challenges.The key to guaranteeing an accurate assessment is accurate detection of different skin features. We present an automatic facial skin feature detection method that works across a variety of skin tones and age groups for selfies in the wild. To be specific, we annotate the locations of acne, pigmentation, and wrinkle for selfie images with different skin tone colors, severity levels, and lighting conditions. The annotation is conducted in a two-phase scheme with the help of a dermatologist to train volunteers for annotation. We employ Unet++ as the network architecture for feature detection. This work shows that the two-phase annotation scheme can robustly detect the accurate locations of acne, pigmentation, and wrinkle for selfie images with different ethnicities, skin tone colors, severity levels, age groups, and lighting conditions.
Low-light image enhancement (LLE) remains challenging due to the unfavorable prevailing low-contrast and weak-visibility problems of single RGB images. In this paper, we respond to the intriguing learning-related question -- if leveraging both accessible unpaired over/underexposed images and high-level semantic guidance, can improve the performance of cutting-edge LLE models? Here, we propose an effective semantically contrastive learning paradigm for LLE (namely SCL-LLE). Beyond the existing LLE wisdom, it casts the image enhancement task as multi-task joint learning, where LLE is converted into three constraints of contrastive learning, semantic brightness consistency, and feature preservation for simultaneously ensuring the exposure, texture, and color consistency. SCL-LLE allows the LLE model to learn from unpaired positives (normal-light)/negatives (over/underexposed), and enables it to interact with the scene semantics to regularize the image enhancement network, yet the interaction of high-level semantic knowledge and the low-level signal prior is seldom investigated in previous methods. Training on readily available open data, extensive experiments demonstrate that our method surpasses the state-of-the-arts LLE models over six independent cross-scenes datasets. Moreover, SCL-LLE's potential to benefit the downstream semantic segmentation under extremely dark conditions is discussed. Source Code: https://github.com/LingLIx/SCL-LLE.
With AlphaGo defeats top human players, reinforcement learning(RL) algorithms have gradually become the code-base of building stronger artificial intelligence(AI). The RL algorithm design firstly needs to adapt to the specific environment, so the designed environment guides the rapid and profound development of RL algorithms. However, the existing environments, which can be divided into real world games and customized toy environments, have obvious shortcomings. For real world games, it is designed for human entertainment, and too much difficult for most of RL researchers. For customized toy environments, there is no widely accepted unified evaluation standard for all RL algorithms. Therefore, we introduce the first virtual user-friendly environment framework for RL. In this framework, the environment can be easily configured to realize all kinds of RL tasks in the mainstream research. Then all the mainstream state-of-the-art(SOTA) RL algorithms can be conveniently evaluated and compared. Therefore, our contributions mainly includes the following aspects: 1.single configured environment for all classification of SOTA RL algorithms; 2.combined environment of more than one classification RL algorithms; 3.the evaluation standard for all kinds of RL algorithms. With all these efforts, a possibility for breeding an AI with capability of general competency in a variety of tasks is provided, and maybe it will open up a new chapter for AI.
Synchrotron radiation sources are widely used in various fields, among which computed tomography (CT) is one of the most important. The amount of effort expended by the operator varies depending on the subject. If the number of angles needed to be used can be greatly reduced under the condition of similar imaging effects, the working time and workload of the experimentalists will be greatly reduced. However, decreasing the sampling angle can produce serious artifacts and blur the details. We try to use a deep learning model which can build high quality reconstruction sparse data sampling from the angle of the image and ResAttUnet are put forward. ResAttUnet is roughly a symmetrical U-shaped network that incorporates similar mechanisms to ResNet and attention. In addition, the mixed precision is adopted to reduce the demand for video memory of the model and training time.
Concealed object detection in Terahertz imaging is an urgent need for public security and counter-terrorism. In this paper, we provide a public dataset for evaluating multi-object detection algorithms in active Terahertz imaging resolution 5 mm by 5 mm. To the best of our knowledge, this is the first public Terahertz imaging dataset prepared to evaluate object detection algorithms. Object detection on this dataset is much more difficult than on those standard public object detection datasets due to its inferior imaging quality. Facing the problem of imbalanced samples in object detection and hard training samples, we evaluate four popular detectors: YOLOv3, YOLOv4, FRCN-OHEM, and RetinaNet on this dataset. Experimental results indicate that the RetinaNet achieves the highest mAP. In addition, we demonstrate that hiding objects in different parts of the human body affect detection accuracy. The dataset is available at https://github.com/LingLIx/THz_Dataset.
A small change of design semantics may affect a user's satisfaction with a product. To modify a design semantic of a given product from personalised brain activity via adversarial learning, in this work, we propose a deep generative transformation model to modify product semantics from the brain signal. We attempt to accomplish such synthesis: 1) synthesising the product image with new features corresponding to EEG signal; 2) maintaining the other image features that irrelevant to EEG signal. We leverage the idea of StarGAN and the model is designed to synthesise products with preferred design semantics (colour & shape) via adversarial learning from brain activity, and is applied with a case study to generate shoes with different design semantics from recorded EEG signals. To verify our proposed cognitive transformation model, a case study has been presented. The results work as a proof-of-concept that our framework has the potential to synthesis product semantic from brain activity.
Automated analysis of mouse behaviours is crucial for many applications in neuroscience. However, quantifying mouse behaviours from videos or images remains a challenging problem, where pose estimation plays an important role in describing mouse behaviours. Although deep learning based methods have made promising advances in mouse or other animal pose estimation, they cannot properly handle complicated scenarios (e.g., occlusions, invisible keypoints, and abnormal poses). Particularly, since mouse body is highly deformable, it is a big challenge to accurately locate different keypoints on the mouse body. In this paper, we propose a novel hourglass network based model, namely Graphical Model based Structured Context Enhancement Network (GM-SCENet) where two effective modules, i.e., Structured Context Mixer (SCM) and Cascaded Multi-Level Supervision module (CMLS) are designed. The SCM can adaptively learn and enhance the proposed structured context information of each mouse part by a novel graphical model with close consideration on the difference between body parts. Then, the CMLS module is designed to jointly train the proposed SCM and the hourglass network by generating multi-level information, which increases the robustness of the whole network. Based on the multi-level predictions from the SCM and the CMLS module, we also propose an inference method to enhance the localization results. Finally, we evaluate our proposed approach against several baselines...
Home-cage social behaviour analysis of mice is an invaluable tool to assess therapeutic efficacy of neurodegenerative diseases. Despite tremendous efforts made within the research community, single-camera video recordings are mainly used for such analysis. Because of the potential to create rich descriptions of mouse social behaviors, the use of multi-view video recordings for rodent observations is increasingly receiving much attention. However, identifying social behaviours from various views is still challenging due to the lack of correspondence across data sources. To address this problem, we here propose a novel multiview latent-attention and dynamic discriminative model that jointly learns view-specific and view-shared sub-structures, where the former captures unique dynamics of each view whilst the latter encodes the interaction between the views. Furthermore, a novel multi-view latent-attention variational autoencoder model is introduced in learning the acquired features, enabling us to learn discriminative features in each view. Experimental results on the standard CRMI13 and our multi-view Parkinson's Disease Mouse Behaviour (PDMB) datasets demonstrate that our model outperforms the other state of the arts technologies and effectively deals with the imbalanced data problem.