Semantic Segmentation On S3Dis Area5
评估指标
Number of params
mAcc
mIoU
oAcc
评测结果
各个模型在此基准测试上的表现结果
比较表格
| 模型名称 | Number of params | mAcc | mIoU | oAcc |
|---|---|---|---|---|
| self-positioning-point-based-transformer-for | N/A | 76.4 | 70.8 | 90.7 |
| point-is-a-vector-a-feature-representation-in | - | 78.1 | 72.3 | 91 |
| dino-in-the-room-leveraging-2d-foundation | - | - | 74.1 | - |
| kpconv-flexible-and-deformable-convolution | 14.1M | 72.8 | 67.1 | - |
| pointmixer-mlp-mixer-for-point-cloud | 6.5M | 77.4 | 71.4 | - |
| tangent-convolutions-for-dense-prediction-in | N/A | 62.2 | - | - |
| point-cloud-oversegmentation-with-graph | 290K | 68.2 | 61.7 | 87.9 |
| dilated-point-convolutions-on-the-receptive | N/A | - | 61.28 | - |
| 190910469 | N/A | 68.3 | 61.85 | 87.18 |
| segcloud-semantic-segmentation-of-3d-point | N/A | 57.4 | 48.9 | - |
| window-normalization-enhancing-point-cloud | N/A | 77.9 | 71.4 | 91.1 |
| scalable-3d-panoptic-segmentation-with | 0.21 | - | 68.1 | - |
| swin3d-a-pretrained-transformer-backbone-for | N/A | 80.5 | 74.5 | 92.7 |
| point-transformer-1 | N/A | - | 41.1 | - |
| subspace-prototype-guidance-for-mitigating | - | 79.5 | 73.3 | 91.9 |
| mamba24-8d-enhancing-global-interaction-in | - | - | 73.5 | - |
| condaformer-disassembled-transformer-with-1 | - | 78.9 | 73.5 | 92.4 |
| serialized-point-mamba-a-serialized-point | - | - | 70.6 | - |
| scf-net-learning-spatial-contextual-features | N/A | 71.8 | 63.7 | 87.2 |
| efficient-3d-semantic-segmentation-with-1 | 212K | 77.3 | 68.9 | 89.5 |
| deep-fusionnet-for-point-cloud-semantic | N/A | 72.3 | 67.2 | - |
| anisotropic-separable-set-abstraction-for | N/A | - | 66.8 | - |
| window-normalization-enhancing-point-cloud | N/A | 78.2 | 72.2 | 91.4 |
| towards-large-scale-3d-representation | N/A | 78.2 | 72.7 | 91.5 |
| surface-representation-for-point-clouds | 0.97M | 76.0 | 68.9 | 90.2 |
| point-transformer-v3-simpler-faster-stronger | - | 80.1 | 74.7 | 92.0 |
| a-large-scale-network-construction-and | 31.2M | 78.7 | 73.5 | 92.0 |
| stratified-transformer-for-3d-point-cloud | 8.0M | 78.1 | 72.0 | 91.5 |
| kpconvx-modernizing-kernel-point-convolution | - | 78.7 | 73.5 | 91.7 |
| deep-parametric-continuous-convolutional-1 | N/A | 67.0 | - | - |
| meta-architecure-for-point-cloud-analysis | N/A | - | 71.3±0.7 | - |
| pointnet-deep-learning-on-point-sets-for-3d | N/A | 49.0 | - | - |
| patchformer-a-versatile-3d-transformer-based | N/A | - | 67.3 | - |
| point-transformer-v2-grouped-vector-attention | N/A | 78.0 | 72.6 | 91.6 |
| beyond-local-patches-preserving-global-local | - | 80.2 | 73.6 | 93.0 |
| exploring-data-efficient-3d-scene | - | - | 72.2 | - |
| semantic-segmentation-for-real-point-cloud | N/A | 73.1 | 65.4 | 88.9 |
| point-transformer-1 | N/A | - | 57.3 | - |
| contrastive-boundary-learning-for-point-cloud | N/A | 77.9 | 71.6 | 91.2 |
| interpretable-edge-enhancement-and | 8.1M | 78.1 | 72.2 | 91.3 |
| pointnext-revisiting-pointnet-with-improved | 41.6M | 77.2 | 71.1 | 91.0 |
| decoupled-local-aggregation-for-point-cloud | 7.0M | 80.0 | 74.1 | 92.2 |
| 模型 43 | - | - | 72.6 | - |
| exploiting-local-geometry-for-feature-and | - | - | 69.4 | 89.4 |
| virtual-multi-view-fusion-for-3d-semantic | N/A | - | 65.38 | - |
| pointhr-exploring-high-resolution | - | 78.7 | 73.2 | 91.8 |
| sonata-self-supervised-learning-of-reliable | - | 81.6 | 76.0 | 93.0 |
| deepgcns-making-gcns-go-as-deep-as-cnns | N/A | - | 52.49 | - |
| lcpformer-towards-effective-3d-point-cloud | N/A | 76.8 | 70.2 | 90.8 |
| point-transformer-1 | 7.8M | 76.5 | 70.4 | 90.8 |
| sat-size-aware-transformer-for-3d-point-cloud | N/A | 78.8 | 72.6 | - |
| 4d-spatio-temporal-convnets-minkowski | 37.9M | 71.7 | 65.4 | - |
| omnivec-learning-robust-representations-with | - | - | 75.9 | - |
| cga-net-category-guided-aggregation-for-point | N/A | - | 68.6 | - |
| ponderv2-pave-the-way-for-3d-foundataion | - | 79.0 | 73.2 | 92.2 |
| mugnet-multi-resolution-graph-neural-network | N/A | - | 63.5 | 88.1 |
| pointcnn-convolution-on-x-transformed-points | N/A | - | - | 85.9 |
| pointweb-enhancing-local-neighborhood | N/A | - | - | 87.0 |
| ckconv-learning-feature-voxelization-for | - | 75.1 | 67.7 | 89.6 |
| large-scale-point-cloud-semantic-segmentation | 280K | 66.5 | 58.04 | 86.38 |