While industrial anomaly detection (AD) technology has experienced notable advancements in recent years, the challenges of generating realistic anomalies and acquiring a deep understanding of normal patterns persist. In this research, we introduce FractalAD, an end-to-end industrial anomaly detection method. The training dataset is curated by synthesizing fractal images and patches derived from normal samples. This unique fractal anomaly generation method is meticulously crafted to capture the intricate morphology of anomalies comprehensively. Furthermore, we have devised a knowledge distillation structure, aimed at extracting inherent knowledge present in normal samples. To bridge the gap between a teacher and a student model, discrepancies are translated into anomaly attention through a cosine similarity attention module. Notably, our proposed method facilitates the utilization of an end-to-end semantic segmentation network for anomaly detection, seamlessly integrating into the existing framework without introducing additional trainable parameters to the backbone and segmentation head. It exhibits clear advantages over alternative methods in terms of both training efficiency and inference speed. The outcomes of ablation studies underscore the efficacy of both fractal anomaly generation and backbone knowledge distillation. Performance experiments substantiate the competitive standing of FractalAD, showcasing commendable results on the MVTec AD dataset and MVTec 3D-AD dataset when compared to other state-of-the-art anomaly detection methods.

Fast and accurate road damage detection is essential for the automatization of road inspection. This paper describes our solution submitted to the Global Road Damage Detection Challenge of the 2020 IEEE International Conference on Big Data, for typical road damage detection in digital images based on deep learning. The recently proposed YOLOv4 is chosen as the baseline network, while the effects of data augmentation, transfer learning, Optimized Anchors, and their combination are evaluated. We propose a novel road damage data generation method based on a generative adversarial network, which can generate multi-class samples with a single model. The evaluation results demonstrate the effectiveness of different tricks and their combinations on the road damage detection task, which provides a reference for practical application. The code of our solution is available at https://github.com/ZhangXG001/RoadDamgeDetection.git.

In recent years, sequential learning has been of great interest due to the advance of deep learning with applications in time-series forecasting, natural language processing, and speech recognition. Recurrent neural networks (RNNs) have achieved superior performance in single-view and synchronous multi-view sequential learning comparing to traditional machine learning models. However, the method remains less explored in asynchronous multi-view sequential learning, and the unalignment nature of multiple sequences poses a great challenge to learn the inter-view interactions. We develop an AMANet (Attention and Memory-Augmented Networks) architecture by integrating both attention and memory to solve asynchronous multi-view learning problem in general, and we focus on experiments in dual-view sequences in this paper. Self-attention and inter-attention are employed to capture intra-view interaction and inter-view interaction, respectively. History attention memory is designed to store the historical information of a specific object, which serves as local knowledge storage. Dynamic external memory is used to store global knowledge for each view. We evaluate our model in three tasks: medication recommendation from a patient's medical records, diagnosis-related group (DRG) classification from a hospital record, and invoice fraud detection through a company's taxation behaviors. The results demonstrate that our model outperforms all baselines and other state-of-the-art models in all tasks. Moreover, the ablation study of our model indicates that the inter-attention mechanism plays a key role in the model and it can boost the predictive power by effectively capturing the inter-view interactions from asynchronous views.

Today, exoskeletons are widely applied to provide walking assistance for patients with lower limb motor incapacity. Most existing exoskeletons are under-actuated, resulting in a series of problems, e.g., interference and unnatural gait during walking. In this study, we propose a novel intelligent autonomous lower extremity exoskeleton (Auto-LEE), aiming at improving the user experience of wearable walking aids and extending their application range. Unlike traditional exoskeletons, Auto-LEE has 10 degrees of freedom, and all the joints are actuated independently by direct current motors, which allows the robot to maintain balance in aiding walking without extra support. The new exoskeleton is designed and developed with a modular structure concept and multi-modal human-robot interfaces are considered in the control system. To validate the ability of self-balancing bipedal walking, three general algorithms for generating walking patterns are researched, and a preliminary experiment is implemented.