Search

With the widespread deployment of surveillance cameras, automatic violence detection has attracted extensive attention from industry and academia. Though researchers have made great progress in video-based violence detection, it is still a challenging task to realize accurate violence detection in real time, especially with limited computing resources. In this paper, we propose a lightweight 2D CNN-based violence detection scheme, which takes advantage of frame-grouping to reduce data redundancy greatly and, meanwhile, enable short-term temporal modeling. In particular, a lightweight 2D CNN, named improved EfficientNet-B0, is constructed by integrating our proposed bi-directional long-term motion attention (Bi-LTMA) module and a temporal shift module (TSM) into the original EfficientNet-B0. The Bi-LTMA takes both spatial and channel dimensions into consideration and captures motion features in both forward and backward directions. The TSM is adopted to realize temporal feature interaction. Moreover, an auxiliary classifier is designed and employed to improve the classification capability and generalization performance of the proposed model. Experiment results demonstrate that the computational cost of the proposed model is 1.21 GFLOPS. Moreover, the proposed scheme achieves accuracies of 100%, 98.5%, 91.67%, and 90.25% on the Movie Fight dataset, the Hockey Fight dataset, the Surveillance Camera dataset, and the RWF-2000 dataset, respectively.

Long-term erosion of granite—a type of hard rock—by hydrothermal fluids and tectonic movement can lead to a fragmentation of the internal structure of the original rock, transfer/replacement of mineral elements, and alteration of the rock’s basic properties. Such changes can be problematic for the construction of water conservancy, hydropower, and road projects. This study adopted the altered Variscan granite in a water diversion project area at the northern foot of the Tianshan Mountains (China) as the research object, and explored the alteration mechanism using thin section identification, X-ray diffraction, major element analysis, and electron probe and oxygen isotope tests. Results showed that the lithology of the granite in the study area is mainly biotite granodiorite and biotite monzonitic granite. Reductions in both the K+ content in plagioclase and the K+ and Ti4+ contents in chlorite indicate that the alteration types within the study area are mainly clayization of feldspar minerals and chloritization of biotite. Biotite granodiorite and biotite monzonitic granite both have low δ18O values. The δ18O value of biotite granodiorite decreases with increase in the Loss on Ignition. The low-δ18O-value granodiorite due to an alteration by hydrothermal fluids transformed from glacier meltwater, groundwater, atmospheric precipitation, and magmatic water; whereas the monzonitic granite might be formed by the reinvasion of low-δ18O-value granodiorite formed in the early stage, which is remelted, assimilated, and rebalanced. The research findings provide a reference for similar research on altered granite at the northern foot of the Tianshan Mountains, and also lay a foundation for subsequent research on its physical/mechanical properties and engineering characteristics.

Recently, the Vision Transformer (ViT) model has been used for various computer vision tasks, due to its advantages to extracting long-range features. To better integrate the long-range features useful for classification, the standard ViT adds a class token, in addition to patch tokens. Despite state-of-the-art results on some traditional vision tasks, the ViT model typically requires large datasets for supervised training, and thus, it still face challenges in areas where it is difficult to build large datasets, such as medical image analysis. In the ViT model, only the output corresponding to the class token is fed to a Multi-Layer Perceptron (MLP) head for classification, and the outputs corresponding to the patch tokens are exposed. In this paper, we propose an improved ViT architecture (called ViT-Patch), which adds a shared MLP head to the output of each patch token to balance the feature learning on the class and patch tokens. In addition to the primary task, which uses the output of the class token to discriminate whether the image is malignant, a secondary task is introduced, which uses the output of each patch token to determine whether the patch overlaps with the tumor area. More interestingly, due to the correlation between the primary and secondary tasks, the supervisory information added to the patch tokens help with improving the performance of the primary task on the class token. The introduction of secondary supervision information also improves the attention interaction among the class and patch tokens. And by this way, ViT reduces the demand on dataset size. The proposed ViT-Patch is validated on a publicly available dataset, and the experimental results show its effectiveness for both malignant identification and tumor localization.

The evolution of interfacial morphology and shear strengths of the joints soldered with Sn-0.3Ag-0.7Cu (SAC0307) and SAC0307-0.05Pr aged at 150 °C for different times (h; up to 840 h) were investigated. The experiments showed the electronic joint soldered with SAC0307-0.05Pr has a much higher shear strength than that soldered with SAC0307 after each period of the aging process. This contributes to the doping of Pr atoms, “vitamins in alloys”, which tend to be adsorbed on the grain surface of interfacial Cu6Sn5 IMCs, inhibiting the growth of IMCs. Theoretical analysis indicates that doping 0.05 wt.% Pr can evidently lower the growth constant of Cu6Sn5 (DCu6), while the growth constant of Cu3Sn (DCu3) decreased slightly. In addition, the electronic joint soldered with SAC0307-0.05Pr still has better ductility than that soldered with SAC0307, even after a 840-h aging process.

The permanent magnet (PM) brushless DC (BLDC) machine has been widely used in various applications due to its high reliability and simplicity. In aerospace applications, the requirements of its dynamic acceleration characteristic and fault protection capability have further increased. This paper proposes an optimized inverter topology composed of a boost converter, two protective power switches, and a three-phase full bridge for the dynamic response improvement and fault protection capability enhancement of PM BLDC machines. The corresponding drive logic, control method, and protection strategy are designed to develop a drive system for an aeronautical fuel pump. The response performance is significantly improved in handling under-voltage and over-rated conditions by utilizing the boost converter during different processes, and the protection capability is enhanced to deal with over-voltage and over-current fault isolation. A boost converter-fed BLDC pump system prototype is built and tested. It is concluded that the simulation and experiment results verify the rationality and validity of the proposed method.