Research map / Detection and segmentation / 3D detection and driving
H^2A^2: Homogeneity-Aware and Heterogeneity-Aware Feature Perception for Unified Indoor 3D Object Detection
Tao Xie, Tao An, Feng Liu, Wensheng Jin, Zhengyu Li, Lijun Zhao, Ruifeng Li
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 SF-KS (structural-feature-aware kernel selection) (ablation) | TR3D-based backbone · ScanNet v2 | mAP@0.5: 54.6 -> 61.3 | Improved |
| add SF-KS (ablation) | TR3D-based backbone · S3DIS | mAP@0.5: 56.8 -> 56.0 | Hurt |
| add NGH (norm-based gradient homogenization) (ablation) | TR3D-based backbone · ScanNet v2 | mAP@0.25: 70.8 -> 72.9 | Improved |
| add NGH on top of SF-KS (ablation) | H2A2 · ScanNet v2 | mAP@0.5: 61.3 -> 63.8 | Improved |
| add NGH on top of SF-KS (ablation) | H2A2 · SUN RGB-D | mAP@0.5: 48.1 -> 51.4 | Improved |
| add task-aware linear modulation (TLM) (ablation) | H2A2 · ScanNet v2 | mAP@0.25: 76.7 -> 77.5 | Improved |
| use TLM instead of linear attention (ablation) | H2A2 · S3DIS | mAP@0.5: 58.8 -> 59.3 | Improved |
| port SF-KS and NGH to FCAF3D (ablation) | FCAF3D · ScanNet v2 | mAP@0.5: 57.3 -> 60.6 | Improved |
| port SF-KS and NGH to FCAF3D (ablation) | FCAF3D · S3DIS | mAP@0.25: 66.7 -> 70.1 | Improved |
| use H2A2 instead of TR3D for zero-shot transfer | TR3D trained on ScanNet · 3RScan zero-shot | mAP@0.25: 47.6 -> 50.7 | Improved |
| add H2A2 modules to TR3D | TR3D · ScanNet v2 | FPS: 23.7 -> 18.2 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- SPGroup3D — SPGroup3D: Superpoint Grouping Network for Indoor 3D Object Detection (AAAI 2024) · 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
- Uncertainty Meets Diversity: A Comprehensive Active Learning Framework for Indoor 3D Object Detection (CVPR 2025 · similarity 0.88)
- Cubify Anything: Scaling Indoor 3D Object Detection (CVPR 2025 · similarity 0.87)
- Rooms from Motion: Un-posed Indoor 3D Object Detection as Localization and Mapping (NeurIPS 2025 · similarity 0.85)
- UniDet3D: Multi-dataset Indoor 3D Object Detection (AAAI 2025 · similarity 0.85)
- 3DiffTection: 3D Object Detection with Geometry-Aware Diffusion Features (CVPR 2024 · similarity 0.85)
- MVSDet: Multi-View Indoor 3D Object Detection via Efficient Plane Sweeps (NeurIPS 2024 · similarity 0.85)
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.
