yolov5d mAP
时间: 2023-09-18 07:13:16 浏览: 174
yolov5.apk
根据引用中的信息,提出的YOLOv4 - 5d的 mAP提高了4.23%。而根据引用中的信息,YOLOV4-5D的检测头由一个3 × 3卷积层和一个1 × 1卷积层组成,每个最终预测的输出通道为3(K 5),其中K为类数。每个位置都与三个不同的anchor相关联。另外,根据引用中的信息,计算YOLOv5d的mAP有两种方法,其中一种是使用faster rcnn中的voc_eval.py进行计算。<span class="em">1</span><span class="em">2</span><span class="em">3</span>
#### 引用[.reference_title]
- *1* *3* [YOLOv4-5D:一种高效的自动驾驶物体检测器](https://blog.csdn.net/Yong_Qi2015/article/details/122757596)[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_2"}}] [.reference_item style="max-width: 50%"]
- *2* [YOLOv3计算recall和mAP](https://blog.csdn.net/sihaiyinan/article/details/87903923)[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_2"}}] [.reference_item style="max-width: 50%"]
[ .reference_list ]
阅读全文