Hyperspectral image-anomaly detection (HSI-AD) has become one of the research hotspots in the field of remote sensing. Because HSI’s features of integrating image and spectrum provide a considerable data basis for abnormal object detection, HSI-AD has a huge application potential in HSI analysis. It is difficult to effectively extract a large number of nonlinear features contained in HSI data using traditional machine learning methods, and deep learning has incomparable advantages in the extraction of nonlinear features. Therefore, deep learning has been widely used in HSI-AD and has shown excellent performance. This review systematically summarizes the related reference of HSI-AD based on deep learning and classifies the corresponding methods into performance comparisons. Specifically, we first introduce the characteristics of HSI-AD and the challenges faced by traditional methods and introduce the advantages of deep learning in dealing with these problems. Then, we systematically review and classify the corresponding methods of HSI-AD. Finally, the performance of the HSI-AD method based on deep learning is compared on several mainstream data sets, and the existing challenges are summarized. The main purpose of this article is to give a more comprehensive overview of the HSI-AD method to provide a reference for future research work.

The complementarity of multimodal signal is essential for video anomaly detection. However, existing methods either lack exploration to multimodal data or ignore the implicit alignment of multimodal features. In our work, we address this problem using a novel fusion method and propose a Multimodal Supervise-Attention enhanced Fusion (MSAF) framework under weak supervision. Our framework can be divided into two parts: 1) the multimodal labels refinement part refines video-level ground truth into pseudo clip-level labels for subsequent training, 2) the multimodal supervise-attention fusion network enhances features via implicitly aligning different information, then fusing them effectively to predict anomaly scores with the help of refined labels. We validate our framework on four challenging datasets: ShanghaiTech, UCF-Crime, LAD, and XD-Violence. Extensive experiments on the benchmarks demonstrate the effectiveness of our framework, which achieves comparable results on several benchmarks and outperforms current state-of-the-art methods on the XD-Violence audiovisual multimodal dataset.

Nonintrusive load monitoring (NILM) is performed by monitoring the total electricity consumption data from the customer’s meter and decomposing it with a load decomposition algorithm to get real-time electricity consumption data for individual appliances. Nonintrusive load decomposition is a single-channel blind source separation problem, which is difficult to implement because of the inherently unidentifiable problem. Recent studies have shown that deep learning is gradually becoming the mainstream approach to solve the NILM problem owing to the availability of data, computing power, and deep network training algorithm models. Among them, the sequence-to-point (seq2point) (Zhang) load decomposition model achieves state-of-the-art prediction because it concentrates on the representation ability of the network at one point. In this article, we optimize the structure of the network model based on the seq2point model, use a temporal convolutional network (TCN) with a more flexible perceptual field of view to train the load decomposition model, and replace the traditional activation function Relu with the activation function GELU. Finally, we test the algorithm in the public dataset UK-DALE and select the mean absolute error and F-score as the main evaluation indexes of the algorithm performance. Comparing the decomposition results of the two algorithms for the same period dataset, the results show that the temporal convolutional network model has substantially improved the performance index of load decomposition while reducing the model parameters.

In the virtual power plant architecture based on edge computing, short-term photovoltaic power prediction is carried out by using multi-source data of photovoltaic power plants stored in photovoltaic intelligent edge terminal (PVIET), which can achieve efficient dispatching and management of renewable energy by the control and coordination center, thus ensuring the safe and stable operation of the power grid. Therefore, this paper proposes a short-term photovoltaic power prediction method that combines improved grey relation analysis (IGRA), efficient channel attention module (ECANet), and temporal convolutional network (TCN). The IGRA algorithm can fully explore the correlation between predicted days and historical days using meteorological environmental data from multiple sources, while the deep TCN model combined with ECANet can effectively avoid the drawbacks of traditional photovoltaic power prediction models. Finally, using the real case of Alice Spring in Australia, the proposed method (IGRA-ECA-TCN) is compared with other hybrid models in different seasons and weather conditions, and the mean absolute error (MAE), root mean square error (RMSE) and coefficient of determination (R2) are selected as the evaluation indicators of algorithm performance. The results indicate that IGRA-ECA-TCN has stronger adaptability and higher accuracy.

A lightweight rotational object detection algorithm, R-YOLOv5, is proposed to address the limitations of traditional object detection algorithms that do not consider the diversity of vehicle scales in drone images and fail to obtain information on rotation angles. The proposed algorithm incorporated an angle prediction branch and introduced a circular smooth label (CSL) angle classification method to make it suitable for detection scenarios based on rotational boxes. A cascaded Swin Transformer block (STrB) is used to reduce computational complexity during feature fusion in the backbone network, further enhancing semantic information and global perception capabilities for small objects. A feature enhancement attention module (FEAM) is proposed to improve the utilization of detailed information through local feature self-supervision. An adaptive spatial feature fusion structure (ASFF) is introduced, which employs features extracted from different levels of the backbone network to perform multi-scale feature fusion. The experimental results show that the detection accuracy reaches 84.91% on the Drone-Vehicle dataset and 90.23% on the UCAS-AOD remote sensing dataset. The lightweight model has a parameter count of only 2.02 million and can achieve 82.6 FPS for high-resolution images, which is significantly better than existing lightweight models and more suitable for real-time detection of rotating vehicles in dense scenes, making it suitable for deployment on a large majority of embedded platforms.