songwsx's repositories
person_search_demo
利用YOLOv3结合行人重识别模型,实现行人的检测识别,查找特定行人
RFSong-7993
设计的轻量级RFB进行行人检测,AP达到0.7993,参数量仅有3.1MB,200 FPS
RFSong-779
重新设计的RFBNet300,模型参数量只有0.99MB,AP达到0.78,200FPS
steel-detect
数钢筋demo,IOU 0.7 下,AP 90.6。训练只要不到十分钟,可以非常愉快的 玩耍
RFB-Person
利用RFB进行行人检测的基线代码
Ultra-Light-Fast-Generic-Face-Detector-1MB
一款超轻量级通用人脸检测模型,模型参数量1MB,320x240输入下计算量仅90MFlops,适用于边缘计算设备、低算力设备以及PC
person-detection
TensorRT person tracking RFBNet300
product_detection
Working with scale: 2nd place solution to Product Detection in Densely Packed Scenes