AI Paper Atlas

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

ICCV 2023 · Detection and segmentation › Open-set and few-shot detection · approach: Open-set and unknown object detection · 799 citations (OpenAlex)

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 changedWhereMeasuredEffect
increase signal scaling factor from 1.0 to 2.0 (ablation)DiffusionDet · COCO valAP: 45.0 -> 45.8Improved
pad GT boxes with Gaussian random boxes instead of repeating GT (ablation)DiffusionDet · COCO valAP: 44.2 -> 45.8Improved
add both DDIM and box renewal in sampling (ablation)DiffusionDet · COCO val, 3 iterationsAP: 44.1 -> 46.6Improved
increase evaluation boxes from 300 to 4000DiffusionDet · COCO valAP: +1.0 APImproved
clone queries for more evaluation boxesDETR · COCO valAP: 38.8 -> 38.4Hurt
increase iteration steps from 1 to 8 with 100 boxesDiffusionDet · COCO valAP: 41.9 -> 46.1Improved
run decoder stages twice (6x2)Sparse R-CNN · COCO to CrowdHuman zero-shotAP50: 66.6 -> 60.6Hurt
increase iteration steps to 2 (6x2)DiffusionDet · COCO to CrowdHuman zero-shotAP50: 66.6 -> 69.7Improved
iterative evaluation, LVIS vs COCODiffusionDet · COCO vs LVISAP: COCO 45.8 -> 46.6; LVIS 29.4 -> 31.5Improved
match N_train and N_eval (ablation)DiffusionDet · COCO valAP: N_train 100 gives 42.9 vs 41.9 for N_train 500 at N_eval 100Improved
apply federated loss to Faster R-CNN and Cascade R-CNNLVIS baselines · LVIS v1.0AP: about 3 AP gainsImproved
increase iteration steps costs speedDiffusionDet · COCO valFPS: 30 -> 20Hurt

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.

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

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.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy