Inferencing on custom model

[zahid26]

Issue Description
I have trained a v9-s model on custom dataset, training got completed and now i had a custom model .ckpt. How do i load this model and do inferencing on image(s)

Additional Context
please provide step by step instructions.

Future Considerations
Save the full model and load using torch.

[tomyjany]

I believe here you can find the solution:
#165

[zahid26]

@tomyjany , Yes the issue is resolved but after the inferencing the bboxes are not coming properly means bboxes are getting plotted at random places with the label and conf on it. Could you please help me out?

I am exactly following the same code as you mentioned above.

[babulamani95]

Facing same issues. Was able to convert the model and run inference successfully (#165 ) with extremely bad results.
I have double checked the training data is correct, training is happening where loss is reducing and mAP is converging. Trained for 350 epochs, inference using .ckpt and .pt, all wrong results

[liamnavarre7]

I'm facing similar issues that others have listed. I have been able to strip the unnecessary keys on the .pt file, loss is reducing, but when I run an inference I see no bounding box being drawn. Hoping someone has a solution

[Rhroan]

These can help you

#165

#212

#213