Training Yolo doesnt improve precision and Recall (Stuck at 0)

[take2rohit]

I have exactly used codes as mentioned here in examples of yolo. I have tested using football dataset and grapes dataset as used in provided notebooks. I am not getting the inference results.

The code on colab implemented by me can me found HERE
Please look into it.

[abhi-kumar]

Try reducing the learning rate and retrain.

[abhi-kumar]

@take2rohit any updates on this?

[take2rohit]

I will let you know by today

[abhi-kumar]

Sure.

[take2rohit]

Trained even with reduced learning rate lr=1e-5 still no change
Training isn't learning anything
Could you please check the code of This Notebook
Please look into it.

[mihir135]

Same issue here. mAP, precision, and recall stuck at 0 for all the epochs. I tried to train with 50 epochs on custom dataset.

[abhi-kumar]

The issue was with certain network architecture conversions, YoloV3 requires certain changes in config files and monk automated it, but this automation had certain errors which have now been rectified completely

[abhi-kumar]

Closing due to inactivity