tph yolov5

TPH-YOLOv5是一种改进的YOLOv5模型，用于无人机捕获场景目标检测。它在YOLOv5基线代码的基础上进行了修改，专注于VisDrone2021数据集。该模型采用了Transformer编码块来替代YOLOv5原始版本中的卷积块和CSP瓶颈块。此外，TPH-YOLOv5还增加了一个用于微小物体检测的预测头，以处理目标的大尺度方差。这个预测头由低层高分辨率的特征图生成，对微小物体更加敏感，从而提高了对微小物体的检测性能。TPH-YOLOv5还集成了CBAM模块来帮助网络在大区域覆盖的图像中找到感兴趣的区域。通过这些改进，TPH-YOLOv5在VisDrone2021测试挑战数据集上达到了很好的性能，比DPNetV3的性能提高了1.81%。<span class="em">1</span><span class="em">2</span><span class="em">3</span>
#### 引用[.reference_title]
- *1* [TPH-YOLOv5用于无人机捕获场景目标检测](https://download.csdn.net/download/weixin_44911037/86823848)[target="_blank" data-report-click={"spm":"1018.2226.3001.9630","extra":{"utm_source":"vip_chatgpt_common_search_pc_result","utm_medium":"distribute.pc_search_result.none-task-cask-2~all~insert_cask~default-1-null.142^v93^chatsearchT3_1"}}] [.reference_item style="max-width: 33.333333333333336%"]
- *2* [TPH-YOLOv5简述](https://blog.csdn.net/qq_40402444/article/details/121166319)[target="_blank" data-report-click={"spm":"1018.2226.3001.9630","extra":{"utm_source":"vip_chatgpt_common_search_pc_result","utm_medium":"distribute.pc_search_result.none-task-cask-2~all~insert_cask~default-1-null.142^v93^chatsearchT3_1"}}] [.reference_item style="max-width: 33.333333333333336%"]
- *3* [TPH-YOLOv5](https://blog.csdn.net/m0_47405013/article/details/127071563)[target="_blank" data-report-click={"spm":"1018.2226.3001.9630","extra":{"utm_source":"vip_chatgpt_common_search_pc_result","utm_medium":"distribute.pc_search_result.none-task-cask-2~all~insert_cask~default-1-null.142^v93^chatsearchT3_1"}}] [.reference_item style="max-width: 33.333333333333336%"]
[ .reference_list ]