VOC format


I have created my dataset on roboflow and export it to PASCAL VOC format. I successfully trained it on the ultralytics web site and export in onnx, I found out that the onnx model had just 1 output and the jetson inference onnx model used 2 of them.
So i tried to train my dataset with train_ssd.py but when i export it and launch detectnet, it doesn’t recongnize any drones.
Can you help me figure it out why
Thanks very much

Hi @antonin.morcrette, how many images are in your dataset and what accuracy was it trained to? You can run train_ssd.py with the --validate-mean-ap flag and it will compute the per-class accuracies for you. Did you try running detectnet on some test images from your dataset before the camera?

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.