当前位置:网站首页>yolov5训练可视化指标的含义
yolov5训练可视化指标的含义
2022-07-01 08:30:00 【余生的观澜】
背景
在v5的训练过程中,使用wandb是一个很常见可视化工具,它有很多的图表,可以对训练过程中可以提取的评价指标进行可视化,今天做一个整理。
可视化指标
LOSS
loss分为cls_loss, box_loss, obj_loss三部分。
cls_loss用于监督类别分类,计算锚框与对应的标定分类是否正确。
box_loss用于监督检测框的回归,预测框与标定框之间的误差(CIoU)。
obj_loss用于监督grid中是否存在物体,计算网络的置信度。
参考资料:https://blog.csdn.net/u011994454/article/details/119564834
metrics
mAP([email protected]),这是一个对检测能力要求更高的标准。
mAP([email protected]),跟Pascal VOC mAP标准计算方式一致;
mAP([email protected][0.5:0.05:0.95]),需要计算10个IoU阈值下的mAP,然后计算平均值。这个评估指标比仅考虑通用IoU阈值(0.5)评估指标更能体现出模型的精度。
https://www.jianshu.com/p/fd9b1e89f983
边栏推荐
- 串口转WIFI模块通信
- How to prevent the other party from saying that he has no money after winning the lawsuit?
- 截图小妙招
- Leetcode t34: find the first and last positions of elements in a sorted array
- Principle and application of single chip microcomputer - principle of parallel IO port
- Rumtime 1200 upgrade: London upgrade support, pledge function update and more
- XX攻击——反射型 XSS 攻击劫持用户浏览器
- 01 NumPy介绍
- [Yu Yue education] Shandong Vocational College talking about railway reference materials
- 长安链同步节点配置与启动
猜你喜欢
随机推荐
[dynamic planning] p1020 missile interception (variant of the longest increasing subsequence)
MATLAB小技巧(16)矩阵特征向量特征值求解一致性验证--层次分析
Leetcode T34: 在排序数组中查找元素的第一个和最后一个位置
Qt的模型与视图
长安链同步节点配置与启动
There are many problems in sewage treatment, and the automatic control system of pump station is solved in this way
MAVROS发送自定义话题消息给PX4
XX攻击——反射型 XSS 攻击劫持用户浏览器
机动目标跟踪——当前统计模型(CS模型)扩展卡尔曼滤波/无迹卡尔曼滤波 matlab实现
Manually dig XSS vulnerabilities
Airsim radar camera fusion to generate color point cloud
Audio-AudioRecord create(一)
Huawei machine test questions column subscription Guide
Codeforces Round #803 (Div. 2) VP补题
Principle and application of single chip microcomputer - off chip development
Book of quantitative trading - reading notes of the man who conquers the market
Yolov5 advanced 7 target tracking latest environment setup
《微机原理》-绪论
挖财打新股安全吗
串口转WIFI模块通信







