Grid R-CNN

Benchmark Model Rank Results
2d-object-detection-on-sardet-100kGrid RCNN#10box mAP: 48.8
object-detection-on-coco-test-devGrid R-CNN (ResNeXt-101-FPN)#143box mAP: 43.2AP50: 63.0AP75: 46.6APS: 25.1APM: 46.5APL: 55.2
object-detection-on-coco-val2017Grid R-CNN (ResNet-101-FPN)#374box AP: 41.3AP50: 60.3AP75: 44.4APS: 23.4APM: 45.8APL: 54.1
object-detection-on-coco-val2017Grid R-CNN (ResNet-50-FPN)#407box AP: 39.6AP50: 58.3AP75: 42.4APS: 22.6APM: 43.8APL: 51.5