About the drawing mistakes in the RTMDet algorithm detailed explanation document

[Nu1sance]

📚 The doc issue

https://mmyolo.readthedocs.io/en/latest/recommended_topics/algorithm_descriptions/rtmdet_description.html

In the CSPNeXtBlock drawn in this picture, the padding of the ConvModule is wrong and different from the actual code.
The convolution blocks drawn in the figure are k=3, s=1, p=0 and k=5, s=1, p=1 respectively. This cannot guarantee that the size of the feature map remains unchanged, and the code is implemented as k=3 ,s=1,p=1 and k=5,s=1,p=2.

Suggest a potential alternative/fix

No response