For unsupervised anomaly detection and localization, a common approach is learning the distribution of normal samples and then use it as a criterion to identify abnormalities. This article proposes a multilevel features restoration method based on damaged images (MLDFR) for anomaly detection and localization. MLDFR seeks to restore the “normal feature” of the test sample. Specifically, we damage the training samples to generate the corresponding samples, and then design a concurrent feature extractor utilizing convolutional neural network and transformer pretrained on ImageNet to completely represent the multilevel features of samples. Additionally, we fully consider the dependencies among local features over long distances and design a feature restoration module. On the challenging, widely used anomaly detection datase (MVTec-AD), metal parts defect detection dataset (MPDD), and beantech anomaly detection dataset (BTAD) of real-world datasets, MLDFR achieves state-of-the-art anomaly localization performance, as well as image-level detection with virtually flawless score. We further report ablation studies, demonstrating MLDFR's effectiveness and generalizability.

At present, pointer meters are still widely used because of their mechanical stability and electromagnetic immunity, and it is the main trend to use a computer vision-based automatic reading system to replace inefficient manual inspection. Many correction and recognition algorithms have been proposed for the problems of skew, distortion, and uneven illumination in the field-collected meter images. However, the current algorithms generally suffer from poor robustness, enormous training cost, inadequate compensation correction, and poor reading accuracy. This paper first designs a meter image skew-correction algorithm based on binary mask and improved Mask-RCNN for different types of pointer meters, which achieves high accuracy ellipse fitting and reduces the training cost by transfer learning. Furthermore, the low-light enhancement fusion algorithm based on improved Retinex and Fast Adaptive Bilateral Filtering (RBF) is proposed. Finally, the improved ResNet101 is proposed to extract needle features and perform directional regression to achieve fast and high-accuracy readings. The experimental results show that the proposed system in this paper has higher efficiency and better robustness in the image correction process in a complex environment and higher accuracy in the meter reading process.

Trajectory prediction is an important task in autonomous driving and monitoring systems. Most of the existing methods pay little attention to the rapidly changing trajectory information, but how to effectively solve this problem is crucial to ensure pedestrian safety. The Gabor transform has inherent advantages for capturing instantaneously changing information. Therefore, for the first time, we introduce the Gabor transformation idea into pedestrian trajectory prediction and propose the Multi-scale Learnable Gabor Transform Network (MlgtNet), which establishes global and local contextual relationships from multi-dimensional and multi-scale perspectives. The network first uses the Multi-scale Feature Dimension Enhancement Module (MFDEM) ascending dimension trajectory sequence, and uses the Multi-scale Gabor Convolution Module (MGCM) to guide the model to establish the dependence of different distances from different dimensions to model the interrelationship between global/local features at different scales and different step sizes. Finally, the Feature Fusion Module (FFM) processes the multimodal information and fuses it with the multi-scale trajectory features to obtain the trajectory prediction representation in different visual fields. The representation results are then used for secondary fusion to obtain the global prediction results. Experimental results show that MlgtNet achieves state-of-the-art performance with its lightweight model size on the vast majority of widely used trajectory prediction datasets from different perspectives.

This research addresses cross-training policies for a flexible assembly cell from the point of view of humanisation and finance. This policy is to determine which labours should be cross-trained and to assign which labours to which tasks. The first application of improved non-dominated sorting genetic algorithm (NSGA-II) to the bi-objective 0–1 integer programming model with an average satisfaction degree and an average salary function is presented. Three performance metrics, including the number of non-dominated solutions, convergence and diversity are used to evaluate the NSGA-II. The optimal algorithm parameters, including the number of iteration, crossover rate and mutation rate are discussed and derived based on a series of experiments. And the experiment results gained by NSGA-II are compared to MOPSO. Computational study shows that the algorithm NSGA-II is convergent and practical to this problem. A series of computational experiments are conducted to get some insights about how the effects of cross-training are influenced by the factors. The results indicate that with regard to Average Satisfaction Degree, balanced preference structure is better than extreme, and with regard to Average Paid Salary, non-uniform salary structure is better than uniform salary structure. Those insights will provide the right direction for practitioners.

Anomaly detection has a wide range of applications and is especially important in industrial quality inspection. Currently, many top-performing anomaly detection models rely on feature embedding-based methods. However, these methods do not perform well on datasets with large variations in object locations. Reconstruction-based methods use reconstruction errors to detect anomalies without considering positional differences between samples. In this study, a reconstruction-based method using the noise-to-norm paradigm is proposed, which avoids the invariant reconstruction of anomalous regions. Our reconstruction network is based on M-net and incorporates multiscale fusion and residual attention modules to enable end-to-end anomaly detection and localization. Experiments demonstrate that the method is effective in reconstructing anomalous regions into normal patterns and achieving accurate anomaly detection and localization. On the MPDD and VisA datasets, our proposed method achieved more competitive results than the latest methods, and it set a new state-of-the-art standard on the MPDD dataset.