EATFormer: Improving Vision Transformer Inspired by Evolutionary Algorithm

Benchmark Model Rank Results
object-detection-on-coco-val2017EATFormer-Base (Mask R-CNN 3x)#173box AP: 49.0AP50: 70.3AP75: 53.6Params (M): 68
object-detection-on-coco-val2017EATFormer-Small (Mask R-CNN 3x)#199box AP: 47.4AP50: 69.3AP75: 51.9Params (M): 44
object-detection-on-coco-val2017EATFormer-Base (Mask R-CNN 1x)#203box AP: 47.2AP50: 69.4AP75: 52.1Params (M): 68