3D Semantic Segmentation On Semantickitti
평가 지표
test mIoU
평가 결과
이 벤치마크에서 각 모델의 성능 결과
비교 표
| 모델 이름 | test mIoU |
|---|---|
| polarnet-an-improved-grid-representation-for | 57.2% |
| pointnet-deep-hierarchical-feature-learning | 20.1% |
| tangent-convolutions-for-dense-prediction-in | 35.9% |
| semantic-segmentation-for-real-point-cloud | 59.9% |
| splatnet-sparse-lattice-networks-for-point | 18.4% |
| towards-large-scale-3d-representation | - |
| dino-in-the-room-leveraging-2d-foundation | 74.4% |
| talos-enhancing-semantic-scene-completion-via | - |
| tornado-net-multiview-total-variation | 63.1% |
| squeezesegv3-spatially-adaptive-convolution | 55.9% |
| number-adaptive-prototype-learning-for-3d | 61.6% |
| 191111236 | 53.9% |
| kprnet-improving-projection-based-lidar | 63.1% |
| oa-cnns-omni-adaptive-sparse-cnns-for-3d | - |
| 2dpass-2d-priors-assisted-semantic | 72.9% |
| frnet-frustum-range-networks-for-scalable | 73.3% |
| multi-projection-fusion-for-real-time | 55.5% |
| pointnet-deep-learning-on-point-sets-for-3d | 14.6% |
| less-is-more-reducing-task-and-model | - |
| meta-rangeseg-lidar-sequence-semantic | 61.0% |
| point-transformer-v2-grouped-vector-attention | 72.6% |
| latticenet-fast-point-cloud-segmentation | 52.9% |
| rethinking-range-view-representation-for | 73.3% |
| multi-scale-interaction-for-real-time-lidar | 55.2% |
| point-transformer-v3-simpler-faster-stronger | 75.5% |
| squeezesegv2-improved-model-structure-and | 39.7% |
| large-scale-point-cloud-semantic-segmentation | 17.4% |
| a-dataset-for-semantic-segmentation-of-point | 49.9% |
| kpconv-flexible-and-deformable-convolution | 58.8% |
| fg-net-fast-large-scale-lidar-point | 53.8% |
| lsk3dnet-towards-effective-and-efficient-3d | 75.6% |
| rangenet-fast-and-accurate-lidar-semantic | 52.2% |
| uniseg-a-unified-multi-modal-lidar | 75.2% |
| less-is-more-reducing-task-and-model | - |
| searching-efficient-3d-architectures-with | 66.4% |
| spherical-transformer-for-lidar-based-3d | 74.8% |
| sparse-single-sweep-lidar-point-cloud | 66.0% |
| 3d-mininet-learning-a-2d-representation-from | 55.8% |
| gfnet-geometric-flow-network-for-3d-point | 65.4% |
| fps-net-a-convolutional-fusion-network-for | 57.1% |
| cylindrical-and-asymmetrical-3d-convolution | 68.9% |
| squeezeseg-convolutional-neural-nets-with | 29.5% |
| using-a-waffle-iron-for-automotive-point | 70.8% |
| af-2-s3net-attentive-feature-fusion-with | 70.8% |
| salsanext-fast-semantic-segmentation-of-lidar | 59.5% |
| point-to-voxel-knowledge-distillation-for-1 | 71.2% |