Research map / Detection and segmentation / Open-set and few-shot detection
DiffusionDet: Diffusion Model for Object Detection
Shoufa Chen, Peize Sun, Yibing Song, Ping Luo
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 |
|---|---|---|---|
| increase signal scaling factor from 1.0 to 2.0 (ablation) | DiffusionDet · COCO val | AP: 45.0 -> 45.8 | Improved |
| pad GT boxes with Gaussian random boxes instead of repeating GT (ablation) | DiffusionDet · COCO val | AP: 44.2 -> 45.8 | Improved |
| add both DDIM and box renewal in sampling (ablation) | DiffusionDet · COCO val, 3 iterations | AP: 44.1 -> 46.6 | Improved |
| increase evaluation boxes from 300 to 4000 | DiffusionDet · COCO val | AP: +1.0 AP | Improved |
| clone queries for more evaluation boxes | DETR · COCO val | AP: 38.8 -> 38.4 | Hurt |
| increase iteration steps from 1 to 8 with 100 boxes | DiffusionDet · COCO val | AP: 41.9 -> 46.1 | Improved |
| run decoder stages twice (6x2) | Sparse R-CNN · COCO to CrowdHuman zero-shot | AP50: 66.6 -> 60.6 | Hurt |
| increase iteration steps to 2 (6x2) | DiffusionDet · COCO to CrowdHuman zero-shot | AP50: 66.6 -> 69.7 | Improved |
| iterative evaluation, LVIS vs COCO | DiffusionDet · COCO vs LVIS | AP: COCO 45.8 -> 46.6; LVIS 29.4 -> 31.5 | Improved |
| match N_train and N_eval (ablation) | DiffusionDet · COCO val | AP: N_train 100 gives 42.9 vs 41.9 for N_train 500 at N_eval 100 | Improved |
| apply federated loss to Faster R-CNN and Cascade R-CNN | LVIS baselines · LVIS v1.0 | AP: about 3 AP gains | Improved |
| increase iteration steps costs speed | DiffusionDet · COCO val | FPS: 30 -> 20 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 3 of 3 on most shared results.
- Sparse R-CNN — Sparse R-CNN: End-to-End Object Detection With Learnable Proposals (CVPR 2021) · better on 5 of 5 shared results
- Deformable DETR — Deformable DETR: Deformable Transformers for End-to-End Object Detection (ICLR 2021) · better on 1 of 1 shared result
- Cascade R-CNN — Cascade R-CNN: Delving Into High Quality Object Detection (CVPR 2018) · 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
- DiffusionTrack: Diffusion Model for Multi-Object Tracking (AAAI 2024 · similarity 0.93)
- Generalized Diffusion Detector: Mining Robust Features from Diffusion Models for Domain-Generalized Detection (CVPR 2025 · similarity 0.90)
- DiffuBox: Refining 3D Object Detection with Point Diffusion (NeurIPS 2024 · similarity 0.90)
- ReDiffDet: Rotation-equivariant Diffusion Model for Oriented Object Detection (CVPR 2025 · similarity 0.89)
- CamoDiffusion: Camouflaged Object Detection via Conditional Diffusion Models (AAAI 2024 · similarity 0.88)
- InstaGen: Enhancing Object Detection by Training on Synthetic Dataset (CVPR 2024 · 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.
