Skip to content

Latest commit

 

History

History
12 lines (7 loc) · 299 Bytes

README.md

File metadata and controls

12 lines (7 loc) · 299 Bytes

Car_detection

Train using the officially provided YOLOv3 code

avg IOU 0.92
loss < 0.05

some results are provided

In addtion, the offcially YOLOv3 is better than Keras version and Retinanet in PyTorch...

When I trained use YOLOv3 in Keras,loss is difficult to convergence(2.72...)