Multimodal Token Fusion for Vision Transformers

Open paper
Benchmark Model Rank Results
semantic-segmentation-on-llrgbd-syntheticTokenFusion (SegFormer-B2)#4mIoU: 64.75
semantic-segmentation-on-kitti-360TokenFusion (RGB-Depth)#7mIoU: 57.44
semantic-segmentation-on-kitti-360TokenFusion (RGB-LiDAR)#10mIoU: 54.55
semantic-segmentation-on-sun-rgbdTokenFusion (S)#10Mean IoU: 53.0%
semantic-segmentation-on-sun-rgbdTokenFusion (Ti)#20Mean IoU: 51.4%
3d-object-detection-on-sun-rgbd-valTokenFusion#11mAP@0.25: 64.9mAP@0.5: 48.3
semantic-segmentation-on-deliverTokenFusion (RGB-Depth)#12mIoU: 60.25
semantic-segmentation-on-deliverTokenFusion (RGB-LiDAR)#18mIoU: 53.01
semantic-segmentation-on-deliverTokenFusion (RGB-Event)#24mIoU: 45.63
3d-object-detection-on-scannetv2TokenFusion#18mAP@0.5: 54.2mAP@0.25: 70.8
semantic-segmentation-on-nyu-depth-v2TokenFusion (S)#28Mean IoU: 54.2%
semantic-segmentation-on-nyu-depth-v2TokenFusion (Ti)#33Mean IoU: 53.3%

Rank counts only results with a code link.