Objects do not disappear: Video object detection by single-frame object location anticipation

Benchmark Model Rank Results
video-object-detection-on-imagenet-vidOurs (Def. DETR + SwinB)#3MAP: 91.3
video-object-detection-on-imagenet-vidOurs (Def. DETR + R101)#7MAP: 87.9
video-object-detection-on-imagenet-vidOurs (Faster RCNN + R101)#9MAP: 87.2