支持检测的类别:
- person (人体)
- head (没戴安全帽)
- helmet (戴安全帽)
工地安全检测,给出检测框和置信度。
- 提供三个模型:
- 小模型(yolov5s 29.7M)
- 中模型(yolov5m 86.8M)
- 大模型(yolov5l 190.8M)
- 测试图片效果(只显示安全帽检测,过滤了其它类别的显示,具体看代码)
- 测试图片效果
- 测试图片效果
运行成功后,命令行应该看到下面的信息:
[INFO ] - [ class: "helmet", probability: 0.89502, bounds: [x=0.956, y=0.525, width=0.044, height=0.067] class: "helmet", probability: 0.85951, bounds: [x=0.237, y=0.439, width=0.036, height=0.046] class: "helmet", probability: 0.81705, bounds: [x=0.901, y=0.378, width=0.036, height=0.052] class: "helmet", probability: 0.80817, bounds: [x=0.250, y=0.399, width=0.029, height=0.040] class: "helmet", probability: 0.80528, bounds: [x=0.771, y=0.336, width=0.029, height=0.043] ]目录:
http://www.aias.top/
Git地址:https://github.com/mymagicpower/AIAS
https://gitee.com/mymagicpower/AIAS
欢迎分享,转载请注明来源:内存溢出
评论列表(0条)