| 摘要: |
| 针对少精、弱精患者的家用精子检测仪嵌入式部署需求，提出一种基于改进you only look once（YOLO）v8n的轻量化精子检测算法.通过对YOLOv8n模型进行结构优化，在Neck层引入高效多分支尺度特征金字塔网络（EMBSFPN），在提高精度的同时保证了模型的轻量化.在检测头部分采用轻量级共享可变形卷积检测（LSDECD）头替换原来的检测头，大大减少了模型的参数量和运算量.实验结果表明，改进后的算法在精子检测任务上实现了良好的性能，平均精度提高了2.3%，模型运算量减少了26.8%，为嵌入式系统上的精子检测应用提供了一种有效的解决方案. |
| 关键词: 改进you only look once（YOLO）v8 精子检测 轻量化 |
| DOI：10.20192/j.cnki.JSHNU（NS）.2025.02.005 |
| 分类号:TP183 |
| 基金项目: |
|
| Research and embedded implementation of lightweight sperm detection algorithm based on improved YOLOv8n |
|
ZHANG Dingni, ZHUANG Tianhao, LI Chuanjiang
|
|
College of Information, Mechanical and Electrical Engineering, Shanghai Normal University, Shanghai 201418, China
|
| Abstract: |
| For the embedded deployment requirements of home sperm testing devices for patients with oligospermia and asthenospermia， a lightweight sperm detection algorithm was proposed based on an improved you only look once（YOLO）v8n. Fisrtly， by optimizing the structure of the YOLOv8n model， the efficient multi-branch & scale feature pyramid network（EMBSFPN） was introduced in the Neck layer， which improved accuracy while ensuring the model’s lightweight. Secondly， the lightweight shared deformable convolutional detection （LSDECD） head was used to replace the original detection head in the detection head part， which significantly reduced the number of parameters and computational complexity of the model. Experimental results showed that the improved algorithm achieved excellent performance in the sperm detection task， with an average accuracy improvement of 2.3% and a model computational load reduction of 26.8%， providing an effective solution for sperm detection application on embedded system. |
| Key words: improved you only look once（YOLO） v8 sperm detection lightweight |
