Research map / Detection and segmentation / 3D detection and driving
DisARM: Displacement Aware Relation Module for 3D Detection
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 DisARM to VoteNet* | VoteNet* on ScanNet V2 · ScanNet V2 | mAP@0.5: 44.2 -> 49.7 | Helped |
| Add DisARM to GroupFree3D*(w2×, L12, O512) | GroupFree3D*(w2×, L12, O512) · ScanNet V2 | mAP@0.5: 52.6 -> 53.6 | Helped |
| Add DisARM to imVoteNet | imVoteNet · SUN RGB-D val | mAP@0.25: 63.4 -> 65.3 | Helped |
| Use all proposals as relation anchors instead of objectness-filtered feature-space FPS (ablation) | VoteNet+DisARM · ScanNet V2 | mAP@0.25: 66.1 -> 63.3 | Hurt |
| Use nearest 15 proposals as relation anchors instead of objectness-filtered feature-space FPS (ablation) | VoteNet+DisARM · ScanNet V2 | mAP@0.25: 66.1 -> 64.3 | Hurt |
| Select 15 relation anchors at random instead of objectness-filtered feature-space FPS (ablation) | VoteNet+DisARM · ScanNet V2 | mAP@0.25: 66.1 -> 65.0 | Hurt |
| Sample anchors using distance FPS instead of objectness-filtered feature-space FPS (ablation) | VoteNet+DisARM · ScanNet V2 | mAP@0.25: 66.1 -> 65.1 | Hurt |
| Sample anchors using feature FPS without objectness filtering (ablation) | VoteNet+DisARM · ScanNet V2 | mAP@0.25: 66.1 -> 65.3 | Hurt |
| Combine distance FPS and feature FPS for anchor sampling (ablation) | VoteNet+DisARM · ScanNet V2 | mAP@0.25: 66.1 -> 65.0 | Hurt |
| Apply feature FPS then distance FPS (ablation) | VoteNet+DisARM · ScanNet V2 | mAP@0.5: 49.7 -> 48.4 | Hurt |
| Use K-means anchors instead of objectness-filtered feature-space FPS (ablation) | VoteNet+DisARM · ScanNet V2 | mAP@0.25: 66.1 -> 65.2 | Hurt |
| Apply distance FPS to K-means clusters (ablation) | VoteNet+DisARM · ScanNet V2 | mAP@0.5: 49.7 -> 48.7 | Hurt |
Closest papers by meaning
- Point2Seq: Detecting 3D Objects As Sequences (CVPR 2022 · similarity 0.86)
- Relation Networks for Object Detection (CVPR 2018 · similarity 0.86)
- One for All: Multi-Domain Joint Training for Point Cloud Based 3D Object Detection (NeurIPS 2024 · similarity 0.86)
- Object DGCNN: 3D Object Detection using Dynamic Graphs (NeurIPS 2021 · similarity 0.86)
- Relation3D : Enhancing Relation Modeling for Point Cloud Instance Segmentation (CVPR 2025 · similarity 0.86)
- DI-V2X: Learning Domain-Invariant Representation for Vehicle-Infrastructure Collaborative 3D Object Detection (AAAI 2024 · 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.