Closed algoteam5 closed 3 years ago
Thank you, but I am getting overwhelmed by your work. There are so many things. Could you please specify the flow of your work.
For example, First the model run detection by running zero_shot_two_stage
, and then convfc_bbox_semantic_head
is used to create bounding box etc...
Since this repo is based on mmdetection, I suggest you to study the official documentation of mmdetection first, and then you can understand the code smoothly.
I have read your paper. I have some questions regarding Figure 3.