基于改进YOLOv7-tiny算法的学生课堂行为检测

doi:10.3969/j.issn.1006-2475.2025.06.002

摘要/Abstract

摘要： 摘要：针对学生课堂行为检测中目标小而密集，识别效果受各种遮挡、光照干扰的问题，提出一种基于YOLOv7-tiny的学生课堂行为检测改进算法YOLOv7-tiny-SCBD。首先，引入卷积块注意力模块在通道和空间2个维度上提取更多特征信息，改善模型识别效果；其次，通过内容感知特征重组上采样算子增大网络感受野，并依照输入特征图信息动态地进行上采样，以获得更清晰、真实的上采样结果；最后，使用Inner-MPDIoU损失函数更全面、精准评估预测框与真实框的相似度，提高模型检测精度。实验结果表明，本文算法在SCB3-U数据集上mAP@0.5达到了81.6%，相较于原始YOLOv7-tiny算法提升3.1个百分点，参数量为6.19 M，计算量为13.5，每秒检测帧数为85.39，可满足对学生课堂行为的有效检测。

关键词: 关键词：YOLOv7-tiny, 学生课堂行为检测, 注意力模块, 感受野, Inner-MPDIoU

Abstract:
Abstract: In response to the challenges of small and densely packed targets， as well as recognition performance affected by various occlusions and lighting interferences in student classroom behavior detection， this paper proposes an improved algorithm for student classroom behavior detection based on YOLOv7-tiny， termed YOLOv7-tiny-SCBD. Firstly， a convolutional block attention module is introduced to extract more feature information in both channel and spatial dimensions， improving the model's recognition performance. Secondly， by utilizing content-aware feature recombination and upsampling operators， the network's receptive field is enlarged， and upsampling is dynamically performed according to the input feature map information to obtain clearer and more realistic upsampling results. Finally， the Inner-MPDIoU loss function is employed to comprehensively and accurately evaluate the similarity between predicted boxes and ground truth boxes， thereby enhancing the model's detection accuracy. Experimental results demonstrate that the proposed algorithm achieves a mAP@0.5 of 81.6% on the SCB3-U dataset， which is a 3.1 percentage points improvement over the original YOLOv7-tiny algorithm. With 6.19 M parameters， 13.5 calculations， and a detection frame rate of 85.39 frames per second， can effectively detect student classroom behavior.

Key words: Key words: YOLOv7-tiny, student classroom behavior detection, attention module, receptive field, Inner-MPDIoU
CARAFE

中图分类号:

TP391

白佳, 谷林. 基于改进YOLOv7-tiny算法的学生课堂行为检测[J]. 计算机与现代化, 2025, 0(06): 9-15.

BAI Jia, GU Lin. Student Classroom Behavior Detection Based on Improved YOLOv7-tiny Algorithm[J]. Computer and Modernization, 2025, 0(06): 9-15.

参考文献

［1］路卫忠，宋正伟，吴宏杰，等.基于深度学习的人体行为检测方法研究综述［J］. 计算机工程与科学， 2021，43（12）:2206-2215.
［2］ SARABU A， SANTRA A K. Human action recognition in videos using convolution long short-term memory network with spatial-temporal networks［J］. Emerging Science Journal， 2021，5（1）:25-33.
［3］ PANG C， LU X Q， LYN L. Skeleton-based action recognition through contrasting two-stream spatial-temporal networks［J］.IEEE Transactions on Multimedia， 2023，25（1）:8699-8711.
［4］ VARSHNEY N， BAKARIYA B. Deep convolutional neural model for human activities recognition in a sequence of video by combining multiple CNN streams［J］. Multimedia Tools and Applications， 2022，81（29）:42117-42129.
［5］ LIANG C W， LU J， YAN W Q. Human Action Recognition From Digital Videos Based on Deep Learning［C］// Proceedings of the 5th International Coference on Control and Computer Vision. ACM， 2022:150-155.
［6］ HUANG Z， TAO M Y， AN N，et al. ER-C3D: Enhancing R-C3-D Network With Adaptive Shrinkage and Symmetrical Multiscale for Behavior Detection［J］. IEEE Transactions on Computational Social Systems， 2024，11（5）：5997
-6009.
［7］ LI C Y， LI L L， JIANG H L， et al. YOLOv6: A single-stage object detection framework for industrial applications ［J］. arXiv preprint arXiv:2209.02976， 2022.
［8］ WANG C Y， BOCHKOVSKIY A， LIAO H Y M. YOLOv7: Trainable bag-of-freebies sets new state-of-the-art for real-time object detectors［C］// Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. IEEE， 2022: 7464-7475.
［9］ VIKASH S， MUNG C， PRATEEK M. SSD: A unified framework for self-supervised outlier detection［J］. arXiv preprint arXiv:2103.12051， 2021.
［10］ CAI Z W， VASCONCELOS N. Cascade R-CNN: High quality object detection and instance segmentation［J］. IEEE Transactions on Pattern Analysis and Machine Intelligence，2021，43（5）:1483-1498.
［11］ LI J N， LIANG X D， SHEN S M，et al. Scale-aware fast R-CNN for pedestrian detection［J］. IEEE Transactions on Multimedia， 2018，20（4）:985-996.
［12］ HE K M， GEORGIA G， DOLLAR P， et al. Mask R-CNN［C］// Proceedings of the 2017 IEEE International Conference on Computer Vision. IEEE， 2019:9157-9166.
［13］董琪琪，刘剑飞，郝禄国，等. 基于改进SSD算法的学生课堂行为状态识别［J］. 计算机工程与设计， 2021，42（10）:2924-2930.
［14］操丹. 基于目标检测算法的学生课堂行为状态识别研究［D］. 天津:河北工业大学， 2023.
［15］杨颜茜. 基于改进YOLOv8s的学生课堂行为识别研究［J］. 现代计算机， 2024，30（2）:33-38.
［16］ CHEN H H，GUAN J S. Teacher-student behavior recognition in classroom teaching based on improved YOLO-v4 and internet of things technology［J］. Electronics， 2022，11（23）:3998-3998.
［17］ ZHANG Y W， WU Z， CHEN X J， et al. Classroom behavior recognition based on improved YOLOv3［C］// International Conference on Artificial Intelligence and Education. IEEE， 2020:93-97.
［18］廖鹏，刘宸铭，苏航，等. 基于深度学习的学生课堂异常行为检测与分析系统［J］. 电子世界，2018，54（8）: 97-98.
［19］谭斌，杨书焓. 基于Faster R-CNN的学生课堂行为检测算法研究［J］. 现代计算机（专业版）， 2018，17（33）:45-47.
［20］ WOO S， PARK J C， LEE J Y， et al. CBAM: Convolutional block attention module［C］// Proceedings of the European Conference on Computer Vision. ACM， 2018:3-19.
［21］ WANG J Q， CHEN K， XU R， et al. Carafe: Content-aware reassembly Of features［C］// Proceedings of the 2019 IEEE/CVF International Conference on Computer Vision. IEEE， 2019:3007-3016.
［22］ ZHANG H， XU C， ZHANG S J. Inner-IoU: More effective intersection over union loss with auxiliary bounding box［J］. arXiv preprint arXiv:2311.0287， 2023.
［23］ MA S L， XU Y. MPDIoU: A loss for efficient and accurate bounding box regression［J］. arXiv preprint arXiv:2307.076
62， 2023.
［24］ YANG F， WANG T. SCB-Dataset3: A benchmark for detecting student classroom behavior［J］. arXiv preprint arXiv:2310.02522， 2023.
［25］ HU J， SHEN L， SUN G. Squeeze-and-excitation networks.［C］// Proceedings of the 2018 IEEE Conference on Computer Vision and Pattern Recognition. IEEE. 2018:7132-7141.
［26］ WANG Q L， WU B G， ZHU P F， et al. ECA-Net: Efficient channel attention for deep convolutional neural networks［C］// Computer Vision and Pattern Recognition. IEEE， 2020:11531-11539.
［27］ REDMON J， FARHADI A. YOLOv3:An Incremental Improvement［J］. arXiv preprint arXiv:1804.02767， 2018.
［28］ ALEXEY B， WANG C Y， LIAO H Y M. YOLOv4: Optimal Speed and Accuracy of Obj-ect Detection［J］. arXiv preprint arXiv:2004.10934， 2020.
［29］ CAI L F，TANG B， XU Y F， et al. Object detection algorithm based on improved YOLOv5［C］// 5th International Conference on Computer Information Science and Application Technology. 2022:124514H1-124514H7.

[1]	岳有军1, 2, 张远锟1, 赵辉1, 2, 王红君1, 2. 基于多尺度特征与注意力模块的室内场景识别方法[J]. 计算机与现代化, 2024, 0(08): 37-42.
[2]	周成诚, 曾庆军, 杨康, 胡家铭, 韩春伟. 基于高效通道注意力模块的运动想象脑电识别[J]. 计算机与现代化, 2023, 0(12): 19-23.
[3]	刘禅奕, 黄丹, 薛林雁, 王涛, 朱桃, . 改进EfficientNet网络的COVID-19 X光分类[J]. 计算机与现代化, 2023, 0(09): 94-99.
[4]	殷远齐, 徐源, 邢远新. 基于YOLO v4的车辆目标检测算法[J]. 计算机与现代化, 2022, 0(07): 8-14.
[5]	樊一娜,郎波,黄静. 基于非经典感受野动态调整机制的图像表征计算模型[J]. 计算机与现代化, 2015, 0(8): 13-18.