search

AlexeyAB / darknet

YOLOv4 / Scaled-YOLOv4 / YOLO - Neural Networks for Object Detection (Windows and Linux version of Darknet )
http://pjreddie.com/darknet/
Other
21.8k stars 7.97k forks source link

How to calculate model parameters in Yolov4-tiny #7704

Open IgnacioTsai opened 3 years ago

IgnacioTsai commented 3 years ago

Here is my only 1 class yolov4-tiny model screenshots: Screenshot_20210513-120718_Gallery

As the Yolov4 paper in Table 1 said, CSPDarknet53 has 27.6 M. I want to know how to calcualte model parameters like that. I search the issue and got similar question about yolov3-tiny, but I still don't know the calculation of parameter on each layer. https://github.com/AlexeyAB/darknet/issues/6022

And I search the way how to calculate conv and route layer, I got the formula as follows: 1 Conv=filter kernel-width kernel-height * output-channel 2 route = output-channel

And I got the result from the picture I post, it is 6686504, about 6.68M.

Does it right?

Thanks.

ryotaro137 commented 2 years ago

I am struggling with that too. Perhaps that is correct. Please let me know what you find out. Thank you.