what does the output shapes (1, None, 6) in the .pb model indicate?

TNTWEN / OpenVINO-YOLO-Automatic-Generation

MIT License

25 stars 1 forks source link

what does the output shapes (1, None, 6) in the .pb model indicate? #7

Closed Abhishek672 closed 2 years ago

Abhishek672 commented 2 years ago

Hey @TNTWEN I have converted the model to .pb using convert_weights_pb.py . I am getting the output_shape as (1, None, 6). The First 4 are the bounding box dimensions. What are the last 2? Is it confidence_scores and class_probabilities? How can I use them in post processing (i.e. Non max suppression)?

Thanks.