Rethinking Transformer-based Set Prediction for Object Detection
                 Zhiqing Sun  Shengcao Cao Yiming Yang        Kris Kitani
                        Carnegie Mellon University
                   {zhiqings,shengcao,yiming,kkitani}@cs.cmu.edu

                Abstract              Transformer [32] to directly output a final set of predictions
                                  without further post-processing. However, it takes extra-
  DETR is a recently proposed Transformer-based m ...