Classroom behavior detection is a key task in constructing intelligent educational environments. However, the existing models are still deficient in detail feature capture capability, multi-layer feature correlation, and multi-scale target adaptability, making it challenging to realize high-precision real-time detection in complex scenes. This paper proposes an improved classroom behavior detection algorithm, YOLO-AMM, to solve these problems. Firstly, we constructed the Adaptive Efficient Feature Fusion (AEFF) module to enhance the fusion of semantic information between different features and improve the model’s ability to capture detailed features. Then, we designed a Multi-dimensional Feature Flow Network (MFFN), which fuses multi-dimensional features and enhances the correlation information between features through the multi-scale feature aggregation module and contextual information diffusion mechanism. Finally, we proposed a Multi-Scale Perception and Fusion Detection Head (MSPF-Head), which significantly improves the adaptability of the head to different scale targets by introducing multi-scale feature perception, feature interaction, and fusion mechanisms. The experimental results showed that compared with the YOLOv8n model, YOLO-AMM improved the mAP0.5 and mAP0.5-0.95 by 3.1% and 4.0%, significantly improving the detection accuracy. Meanwhile, YOLO-AMM increased the detection speed (FPS) by 12.9 frames per second to 169.1 frames per second, which meets the requirement for real-time detection of classroom behavior.

The rising issue of an aging population has intensified the focus on the health concerns of the elderly. Among these concerns, falls have emerged as a predominant health threat for this demographic. The YOLOv5 family represents the forefront of techniques for human fall detection. However, this algorithm, although advanced, grapples with issues such as computational demands, challenges in hardware integration, and vulnerability to occlusions in the designated target group. To address these limitations, we introduce a pioneering lightweight approach named CGNS-YOLO for human fall detection. Our method incorporates both the GSConv module and the GDCN module to reconfigure the neck network of YOLOv5s. The objective behind this modification is to diminish the model size, curtail floating-point computations during feature channel fusion, and bolster feature extraction efficacy, thereby enhancing hardware adaptability. We also integrate a normalization-based attention module (NAM) into the framework, which concentrates on salient fall-related data and deemphasizes less pertinent information. This strategic refinement augments the algorithm’s precision. By embedding the SCYLLA Intersection over Union (SIoU) loss function, our model benefits from faster convergence and heightened detection precision. We evaluated our model using the Multicam dataset and the Le2i Fall Detection dataset. Our findings indicate a 1.2% enhancement in detection accuracy compared with the conventional YOLOv5s framework. Notably, our model realized a 20.3% decrease in parameter tally and a 29.6% drop in floating-point operations. A comprehensive instance analysis and comparative assessments underscore the method’s superiority and efficacy.