Research map / Detection and segmentation / CNN features
CenterNet++ for Object Detection
Kaiwen Duan, Song Bai, Lingxi Xie, Honggang Qi, Qingming Huang, Qi Chuan Tian
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 central region exploration (center keypoint branch) (ablation) | CornerNet511-52 baseline · MS-COCO val | AP: 37.6 -> 39.9 | Improved |
| add central region exploration (center keypoint branch) (ablation) | CornerNet511-52 baseline · MS-COCO val | AP large objects: 52.2 -> 52.3 | No clear effect |
| add center pooling to the central region exploration (ablation) | CenterNet511-52 · MS-COCO val | AP: 39.9 -> 40.8 | Improved |
| add center pooling (ablation) | CenterNet511-52 · MS-COCO val | AP large objects: 52.2 -> 53.6 | Improved |
| replace corner pooling with cascade corner pooling (ablation) | CornerNet511-52 baseline · MS-COCO val | AP: 37.6 -> 38.3 | Improved |
| replace corner pooling with cascade corner pooling (ablation) | CenterNet511-52 · MS-COCO val | AP: 40.8 -> 41.3 | Improved |
| replace corner pooling with cascade corner pooling (ablation) | CornerNet511-52 baseline · MS-COCO val | AP large objects: 52.2 vs 52.2 | No clear effect |
| use multi-resolution detection layers rather than a single-resolution output (ablation) | MR-CenterNet R-50 · MS-COCO val | AP: 40.9 -> 45.7 | Improved |
| add center keypoint modeling to CornerNet (SR-CenterNet vs CornerNet) | HG-52 · MS-COCO test-dev | AP: 37.8 -> 41.6 | Improved |
| add center keypoint modeling to CornerNet, multi-scale testing | HG-104 · MS-COCO test-dev | AP: 42.1 -> 47.0 | Improved |
| add center keypoint modeling to CornerNet | HG-52/HG-104 · MS-COCO val | average false discovery AF small boxes: decrease of 9.5% by CenterNet511-52 and 9.6% by CenterNet511-104 | Improved |
| replace predicted center keypoints with ground truth | SR-CenterNet HG-52 · MS-COCO val | AP: 41.3 -> 56.5 | Improved |
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.
- CornerNet — CornerNet: Detecting Objects as Paired Keypoints (ECCV 2018) · better on 4 of 4 shared results
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
- CenterNet: Keypoint Triplets for Object Detection (ICCV 2019 · similarity 0.95)
- Bottom-Up Object Detection by Grouping Extreme and Center Points (CVPR 2019 · similarity 0.89)
- CornerNet: Detecting Objects as Paired Keypoints (ECCV 2018 · similarity 0.88)
- Corner Proposal Network for Anchor-free, Two-stage Object Detection (ECCV 2020 · similarity 0.88)
- Finding Your (3D) Center: 3D Object Detection Using a Learned Loss (ECCV 2020 · similarity 0.88)
- CenterFormer: Center-based Transformer for 3D Object Detection (ECCV 2022 · 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.
