rolson24
commented
1 month ago For the YOLOX detector, I’m pretty sure MASA does not use the detector as a feature backbone, the plug and play masa adapter has its own RCNN backbone to get the features directly from the image. So you can use any old detector and just pass the detections and the image into the MASA adapter.
On Thu, Aug 1, 2024 at 7:38 AM docShen @.***> wrote:
Thank you for your outstanding work. May I ask, it seems that the YOLOX-x model mentioned here was not obtained from the YOLOX official website GitHub (because the YOLOX-x weights provided in this project are different from the size of the YOLOX-x model on the YOLOX official website), but I need to retrain this model. How can I retrain YOLOX-x. Thank you！
— Reply to this email directly, view it on GitHub https://github.com/siyuanliii/masa/issues/30, or unsubscribe https://github.com/notifications/unsubscribe-auth/AX2EJPD5H4IMQSILZSITMALZPIMZ3AVCNFSM6AAAAABL2MAOJ2VHI2DSMVQWIX3LMV43ASLTON2WKOZSGQ2DEMRSGY4DONQ . You are receiving this because you are subscribed to this thread.Message ID: @.***>
Thank you for your outstanding work. May I ask, it seems that the YOLOX-x model mentioned here was not obtained from the YOLOX official website GitHub (because the YOLOX-x weights provided in this project are different from the size of the YOLOX-x model on the YOLOX official website), but I need to retrain this model. How can I retrain YOLOX-x.
Thank you！