Research map / Point clouds / Segmentation and detection
Tangent Convolutions for Dense Prediction in 3D
Maxim Tatarchenko, Jaesik Park, Vladlen Koltun, Qian-Yi Zhou
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Use tangent convolutions rather than OctNet | Semantic scene segmentation · ScanNet | mIoU: 18.1 -> 40.9 | Improved |
| Use tangent convolutions rather than PointNet | Semantic scene segmentation · S3DIS | mIoU: 41.3 -> 52.8 | Improved |
| Use tangent convolutions rather than SnapNet | Semantic scene segmentation · Semantic3D | mIoU: 67.7 -> 66.4 | Hurt |
| Add RGB input to distance, height, and normals (ablation) | Tangent convolution network · Semantic3D | mIoU: 52.5 -> 66.4 | Improved |
| Add RGB input to distance, height, and normals (ablation) | Tangent convolution network · S3DIS | mIoU: 51.7 -> 52.8 | Improved |
| Add RGB input to distance, height, and normals (ablation) | Tangent convolution network · ScanNet | mIoU: 40.7 -> 40.9 | No clear effect |
| Add height input to distance (ablation) | Tangent convolution network · Semantic3D | mIoU: 58.1 -> 58.0 | No clear effect |
| Add height input to distance (ablation) | Tangent convolution network · ScanNet | mIoU: 40.9 -> 40.3 | Hurt |
| Add height input to distance (ablation) | Tangent convolution network · S3DIS | mIoU: 49.8 -> 50.0 | No clear effect |
| Add normals to distance and height (ablation) | Tangent convolution network · Semantic3D | mIoU: 58.0 -> 52.5 | Hurt |
| Add normals to distance and height (ablation) | Tangent convolution network · ScanNet | mIoU: 40.3 -> 40.7 | No clear effect |
| Add normals to distance and height (ablation) | Tangent convolution network · S3DIS | mIoU: 50.0 -> 51.7 | Improved |
Closest papers by meaning
- Fully-Convolutional Point Networks for Large-Scale Point Clouds (ECCV 2018 · similarity 0.90)
- PointConv: Deep Convolutional Networks on 3D Point Clouds (CVPR 2019 · similarity 0.88)
- A-CNN: Annularly Convolutional Neural Networks on Point Clouds (CVPR 2019 · similarity 0.88)
- 3D Semantic Segmentation With Submanifold Sparse Convolutional Networks (CVPR 2018 · similarity 0.88)
- PointNet: Deep Learning on Point Sets for 3D Classification and Segmentation (CVPR 2017 · similarity 0.88)
- Towards Semantic Segmentation of Urban-Scale 3D Point Clouds: A Dataset, Benchmarks and Challenges (CVPR 2021 · similarity 0.88)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
