Research map / Point clouds / Segmentation and detection
TACO-Net: Topological Signatures Triumph in 3D Object Classification
Anirban Ghosh, Ayan Dutta
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 |
|---|---|---|---|
| add density/erosion/dilation/signed-distance features to height features (ablation) | TACO-Net · ModelNet40 | overall accuracy: 98.29 -> 98.82 | Improved |
| add radial filtrations to DEDS + height (ablation) | TACO-Net · ModelNet40 | overall accuracy: 98.82 -> 99.05 | Improved |
| add height filtration to DEDS features (ablation) | TACO-Net · ModelNet40 | overall accuracy: 96.52 -> 98.82 | Improved |
| add amplitude features to entropy-only vectors (ablation) | TACO-Net · ModelNet40 | overall accuracy: 96.16 -> 99.05 | Improved |
| add third Conv1D layer (ablation) | TACO-Net 1D CNN · ModelNet40 | overall accuracy: 98.18 -> 99.05 | Improved |
| add second and third Conv1D layers to a single layer (ablation) | TACO-Net 1D CNN · ModelNet40 | overall accuracy: 94.76 -> 99.05 | Improved |
| use voxel size 0.05 instead of 0.03 (ablation) | TACO-Net · ModelNet40 | overall accuracy: 98.61 -> 99.05 | Improved |
| use voxel size 0.05 instead of 0.07 (ablation) | TACO-Net · ModelNet40 | overall accuracy: 98.30 -> 99.05 | Improved |
| replace 1D CNN with 2.2M-parameter transformer (ablation) | TACO-Net · ModelNet40 | overall accuracy: 99.05 -> 62.20 | Hurt |
| replace 1D CNN with XGBoost (ablation) | TACO-Net · ModelNet40 | overall accuracy: 99.05 -> 81.35 | Hurt |
| replace 1D CNN with Random Forest (ablation) | TACO-Net · ModelNet40 | overall accuracy: 99.05 -> 78.35 | Hurt |
| add corrupted-copy augmentation for 20% of training data (ablation) | TACO-Net · ModelNet40-C high rotation | overall accuracy: 49.39 -> 96.11 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- PointMamba — PointMamba: A Simple State Space Model for Point Cloud Analysis (NeurIPS 2024) · better on 1 of 1 shared result
- PointBERT — Point-BERT: Pre-Training 3D Point Cloud Transformers With Masked Point Modeling (CVPR 2022) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Revisiting Point Cloud Classification: A New Benchmark Dataset and Classification Model on Real-World Data (ICCV 2019 · similarity 0.88)
- Deep Hough Voting for 3D Object Detection in Point Clouds (ICCV 2019 · similarity 0.87)
- VoxelNet: End-to-End Learning for Point Cloud Based 3D Object Detection (CVPR 2018 · similarity 0.86)
- Tangent Convolutions for Dense Prediction in 3D (CVPR 2018 · similarity 0.86)
- FILTR: Extracting Topological Features from Pretrained 3D Models (CVPR 2026 · similarity 0.86)
- PointNet: Deep Learning on Point Sets for 3D Classification and Segmentation (CVPR 2017 · similarity 0.86)
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.
