This paper addresses the challenges of detecting students' hand-raising behavior in classrooms, which include small target sizes, complex student features, and distracting classroom background information. We propose an improved YOLOv7-based algorithm, Deformable Simulated-SE Enhanced YOLOv7 (DSS-YOLOv7), specifically designed for classroom behavior detection. The proposed algorithm introduces several enhancements: it optimizes the Feature Pyramid Networks (FPN) structure by incorporating Deformable Convolutions (DCNv2), which effectively capture variations in students' behaviors, thereby improving detection accuracy and robustness. Additionally, the Squeeze-and-Excitation (SE) attention mechanism is introduced to enhance the model's focus on important features and improve recognition accuracy. The integration of the Simple Attention Module (SimAM) helps to reduce model complexity and minimize confusion caused by background information. Furthermore, data augmentation techniques are applied to the dataset, significantly enhancing the model's feature learning capabilities. Experimental results demonstrate that the DSS-YOLOv7 model achieves a 2.6% improvement in mean Average Precision ([email protected]), reaching 85%, compared to YOLOv7 on the SCB-Dataset. The DSS-YOLOv7 model exhibits superior performance in detecting classroom behaviors under complex conditions.

With the rapid development of manufacturing, increasingly stringent requirements for material quality and inspection efficiency have promoted the widespread application of deep learning-based nondestructive testing technologies in industrial quality control. In recent years, steel surface defect detection has expanded from conventional inspection scenarios with controlled imaging conditions, such as steel strips, plates, and welds, to more challenging applications involving the inner surfaces of steel pipes, castings, and complex industrial components. This paper reviews recent advances in deep learning-based steel surface defect detection from an industrial application-oriented perspective, with particular emphasis on developments in application scenarios, datasets, methodological frameworks, and research trends. It analyzes representative reviews and their limitations, summarizes publicly available datasets covering conventional flat steel products, pipes, and castings, examines the defect characteristics of different inspection objects, and reviews the development of deep learning and its industrial applications. Furthermore, defect detection and segmentation methods are systematically organized according to practical requirements, including challenging imaging conditions, limited annotated data, small and low-contrast defects, real-time deployment, and cross-domain generalization. Finally, the key challenges, emerging trends, future research directions, and priorities for the next stage of development are discussed.

With the rising global demand for electricity, transmission infrastructure is becoming increasingly important as a key support for ensuring stable and reliable power supply.In recent years, UAVs have been widely used in the inspection and maintenance of transmission equipment due to their advantages of high efficiency, flexibility and intelligence, which have greatly improved the operation and maintenance efficiency and safety level.However, the transmission equipment itself is exposed to harsh natural environments during prolonged use, such as high temperatures, humidity changes, wind and sand erosion, as well as electromagnetic interference, coupled with complex topographical features, such as mountainous, hilly, and forested areas, which result in the transmission equipment inspection process being challenged by occlusion and large differences in dimensions.To cope with these problems, this paper proposes ACCYolo. a model based on the YOLOv10n architecture with the goal of improving image detection of transmission equipment under multi-scale and occluded targets in UAV-based scenes.On the one hand, the ACCYolo model, to solve the occlusion problem, incorporates the Acmix model, which incorporates the self-attention mechanism to achieve dynamic feature extraction, effectively improving the detection performance of the model in overlapping scenes.On the other hand, in order to cope with the size difference problem in multi-scale detection, the GELAN structure combines a lightweight design with the Programmable Gradient Information (PGI) mechanism to improve the accuracy of multi-scale target detection, while the ASFF module is designed to improve the accuracy of multi-scale target detection through adaptive spatial feature fusion.The experimental results show that. The proposed method shows significant advantages in transmission equipment monitoring tasks, Overall mAP@50 raise to 0.950, and provides an effective program to ensure the reliability of power supply.

Time series clustering plays a vital role in various analytical and pattern recognition tasks by partitioning structurally similar sequences into semantically coherent groups, thereby facilitating downstream analysis. However, building high-quality clustering models remains challenging due to three key issues: (i) capturing dynamic shape variations across sequences, (ii) ensuring discriminative cluster structures, and (iii) enabling end-to-end optimization. To address these challenges, we propose KDiscShapeNet, a structure-aware clustering framework that systematically extends the classical k-Shape model. First, to enhance temporal structure modeling, we adopt Kolmogorov–Arnold Networks (KAN) as the encoder, which leverages high-order functional representations to effectively capture elastic distortions and multi-scale shape features of time series. Second, to improve intra-cluster compactness and inter-cluster separability, we incorporate a dual-loss constraint by combining Center Loss and Supervised Contrastive Loss, thus enhancing the discriminative structure of the embedding space. Third, to overcome the non-differentiability of traditional K-Shape clustering, we introduce Differentiable k-Shape, embedding the normalized cross-correlation (NCC) metric into a differentiable framework that enables joint training of the encoder and the clustering module. We evaluate KDiscShapeNet on nine benchmark datasets from the UCR Archive and the ETT suite, spanning healthcare, industrial monitoring, energy forecasting, and astronomy. On the Trace dataset, it achieves an ARI of 0.916, NMI of 0.927, and Silhouette score of 0.931; on the large-scale ETTh1 dataset, it improves ARI by 5.8% and NMI by 17.4% over the best baseline. Statistical tests confirm the significance of these improvements (p < 0.01). Overall, the results highlight the robustness and practical utility of KDiscShapeNet, offering a novel and interpretable framework for time series clustering.