leggedrobotics/darknet_ros

Error when using my trained network

RodrigoFBernardo opened this issue · 0 comments

I am training my network, with my objects using YOLO v3, when I test it using the darknet source it works fine.

When I use the darknet_ros package my network doesn't work so well anymore. For example, if I use a realsense camera I can't detect any obstacles, but if I use a regular web cam I can detect them. What can cause this?

The realsense camera has higher quality than the web cam so it should work better...