Yolo V4 detects twice to one object

Question

I am training darknet/yolov4 for object detection. When I trained the model and look at the demo, yolov4 detects two objects for one object. How can I fix that problem.

[

Answer 1

One way to fix multiple predictions is to increase the threshold. Try including the -thresh 0.5 in your test command. This will allow not ignore predictions with a score less than 0.5.