Large-scale Point Cloud Semantic Segmentation with Superpoint Graphs

Benchmark Model Rank Results
3d-semantic-segmentation-on-dalesSPG#7mIoU: 60.6Overall Accuracy: 95.5Model size: 280K
3d-semantic-segmentation-on-semantickittiSPGraph#34test mIoU: 17.4%
3d-semantic-segmentation-on-sensaturbanSPGraph#7mIoU: 37.29
semantic-segmentation-on-s3disSPG#39Mean IoU: 62.1mAcc: 73oAcc: 85.5Number of params: 0.290M
semantic-segmentation-on-s3dis-area5SPG#41mIoU: 58.04mAcc: 66.5oAcc: 86.38Number of params: 280K
semantic-segmentation-on-semantic3dSPG#5mIoU: 76.2%oAcc: 92.9%
semantic-segmentation-on-semantic3dSPG#8mIoU: 73.2%