search

jkjung-avt / tensorrt_demos

TensorRT MODNet, YOLOv4, YOLOv3, SSD, MTCNN, and GoogLeNet
https://jkjung-avt.github.io/
MIT License
1.74k stars 545 forks source link

Multiple Bounding Boxes #582

Closed soukaina11111 closed 1 year ago

soukaina11111 commented 1 year ago

hi @jkjung-avt, I’ve trained a model for the detection of person and hat and then convert it, but it is giving multiple bounding box for one detection especially for the class hat like the figure below.

example

NB: I'm using yolov4-csp-swish 640 map 89%, and when I do the inference with opencv I don’t have that problem.

Any idea how to solve it?

Thank you in advance.

jkjung-avt commented 1 year ago

What are CLS0 (class 0) and CLS1 (class 1) of your custom model?

soukaina11111 commented 1 year ago

class 0 : person, class 1: Hat.

jkjung-avt commented 1 year ago

How would you label (ground truth) this picture then?

Or, what is the inference result with OpenCV?

soukaina11111 commented 1 year ago

this is the result i got with opencv test

jkjung-avt commented 1 year ago

Thanks for reporting this issue. But currently I don't have an idea what the root cause is...

soukaina11111 commented 1 year ago

okay thanks for your response.