@TNTWEN
I have converted the model to .pb and then to .tflite. While doing inference I am getting many bounding boxes in the final image.
I am using yolov4-tiny and have written the postprocessing code for yolov4tiny tflite inference.
The prediction by tflite model is (1,2535,6). But for multiple bounding boxes the confidence score is same. Let say 50 boxes has same confidence score 0.45. When I set a threshold during inference, either all bboxes are disappeared, else all bboxes appears.
@TNTWEN I have converted the model to .pb and then to .tflite. While doing inference I am getting many bounding boxes in the final image.
I am using yolov4-tiny and have written the postprocessing code for yolov4tiny tflite inference.
The prediction by tflite model is (1,2535,6). But for multiple bounding boxes the confidence score is same. Let say 50 boxes has same confidence score 0.45. When I set a threshold during inference, either all bboxes are disappeared, else all bboxes appears.
How to fix this issue. Thanks
Thanks