| action-classification-on-toyota-smarthome | I3D + Non Local | #7 | CS: 53.6CV1: 34.3CV2: 43.9 |
| action-recognition-in-videos-on-something-something-v1 | NL I3D | #57 | Top 1 Accuracy: 44.4 |
| instance-segmentation-on-coco-minival | Mask R-CNN (ResNext-152, +1 NL) | #90 | mask AP: 40.3 |
| instance-segmentation-on-coco-minival | Mask R-CNN (ResNet-101, +1 NL) | #106 | mask AP: 37.1 |
| instance-segmentation-on-coco-minival | Mask R-CNN (ResNet-50, +1 NL) | #110 | mask AP: 35.5 |
| human-pose-keypoint-estimation-on-coco-val2017 | Mask R-CNN + NL blocks (4 in head, 1 in backbone) | #101 | AP: 66.5 |
| action-classification-on-kinetics-400 | I3D + NL | #129 | Acc@1: 77.7Acc@5: 93.3 |
| object-detection-on-coco-val2017 | Mask R-CNN (ResNeXt-152 + 1 NL) | #310 | box AP: 45.0AP50: 67.8AP75: 48.9 |
| object-detection-on-coco-val2017 | Mask R-CNN (ResNet-101 + 1 NL) | #385 | box AP: 40.8AP50: 63.1AP75: 44.5 |
| object-detection-on-coco-val2017 | Mask R-CNN (ResNet-50 + 1 NL) | #413 | box AP: 39.0AP50: 61.1AP75: 41.9 |