RMT: Retentive Networks Meet Vision Transformers

Benchmark Model Rank Results
object-detection-on-coco-val2017RMT-B (Cascade Mask R-CNN 3x)#84box AP: 54.5AP50: 72.8AP75: 59.0Params (M): 111
object-detection-on-coco-val2017RMT-S (Cascade Mask R-CNN 3x)#99box AP: 53.2AP50: 72.0AP75: 57.8Params (M): 83
object-detection-on-coco-val2017RMT-B (Mask R-CNN 3x)#117box AP: 52.2AP50: 72.9AP75: 57.0Params (M): 73
object-detection-on-coco-val2017RMT-L (Mask R-CNN 1x)#128box AP: 51.6AP50: 73.1AP75: 56.5Params (M): 114