Used detection model is Cascade Mask-RCNN with Swin-B backbone. get_model ... it already beats the RetinaNet, Mask R-CNN, and Source code for mmdet.
確定! 回上一頁