Research map / Detection and segmentation / 3D detection and driving
Dense RepPoints: Representing Visual Objects with Dense Point Sets
Ze Yang, Yinghao Xu, Han Xue, Zheng Zhang Raquel Urtasun, Liwei Wang, Stephen Lin, Han Hu
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 distance transform sampling instead of boundary sampling at many points (ablation) | Dense RepPoints · COCO val mask AP, n=729 | mask AP: 24.1 -> 33.8 | Improved |
| use distance transform sampling instead of boundary sampling at few points (ablation) | Dense RepPoints · COCO val mask AP, n=9 | mask AP: 19.7 -> 13.9 | Hurt |
| use distance transform sampling instead of grid sampling (ablation) | Dense RepPoints · COCO val mask AP, n=9 | mask AP: 5.0 -> 13.9 | Improved |
| use distance transform sampling instead of grid sampling (ablation) | Dense RepPoints · COCO val mask AP, n=729 | mask AP: 32.8 -> 33.8 | Improved |
| use triangulation instead of concave hull post-processing (ablation) | Dense RepPoints with DTS · COCO val mask AP, n=225 | mask AP: 23.4 -> 32.8 | Improved |
| use set-to-set Chamfer supervision instead of point-to-point (ablation) | Dense RepPoints with DTS · COCO val mask AP, n=9 | mask AP: 10.7 -> 13.9 | Improved |
| use set-to-set Chamfer supervision instead of point-to-point (ablation) | Dense RepPoints with DTS · COCO val mask AP, n=729 | mask AP: 32.6 -> 33.8 | Improved |
| add group pooling and shared offset fields (ablation) | RepPoints · COCO val G FLOPS, n=81 | G FLOPS: 409.46 -> 212.60 | Improved |
| add group pooling and shared offset fields (ablation) | RepPoints · COCO val detection mAP, n=81 | box mAP: 37.5 -> 37.5 | No clear effect |
| increase number of points from 81 to 729 (ablation) | Dense RepPoints · COCO val mask AP | mask AP: 31.5 -> 33.8 | Improved |
| add instance mask supervision (Dense RepPoints) to detection (ablation) | Dense RepPoints · COCO val detection mAP, n=81 | box mAP: 37.5 -> 39.4 | Improved |
| add instance mask branch to Faster R-CNN (Mask R-CNN) (ablation) | Mask R-CNN · COCO val detection mAP | box mAP: 36.4 -> 37.3 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 4 of 4 on most shared results.
- PolarMask — PolarMask: Single Shot Instance Segmentation With Polar Representation (CVPR 2020) · better on 1 of 1 shared result
- SOLO — SOLO: Segmenting Objects by Locations (ECCV 2020) · better on 1 of 1 shared result
- TensorMask — TensorMask: A Foundation for Dense Object Segmentation (ICCV 2019) · better on 1 of 1 shared result
- Faster R-CNN — Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks (TPAMI 2016) · 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
- Continuous Surface Embeddings (NeurIPS 2020 · similarity 0.88)
- RepPoints: Point Set Representation for Object Detection (ICCV 2019 · similarity 0.86)
- CodeSLAM — Learning a Compact, Optimisable Representation for Dense Visual SLAM (CVPR 2018 · similarity 0.86)
- DenseReg: Fully Convolutional Dense Shape Regression In-The-Wild (CVPR 2017 · similarity 0.85)
- Learning Efficient Point Cloud Generation for Dense 3D Object Reconstruction (AAAI 2018 · similarity 0.85)
- DOCTR: Disentangled Object-Centric Transformer for Point Scene Understanding (AAAI 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.
