AI Paper Atlas

Research map / Detection and segmentation / Open-set and few-shot detection

Detecting Objects with Context-Likelihood Graphs and Graph Refinement

Aritra Bhowmik, Yu Wang, Nora Baka, Martin R. Oswald, Cees G. M. Snoek

ICCV 2023 · Detection and segmentation › Open-set and few-shot detection · approach: Visual relationship and scene graph detection · 5 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
add context-likelihood graph built from initial class predictions and relation priors (ablation)Faster-RCNN · VG1000AP: 7.3 -> 8.2Improved
add energy-based graph refinement on top of context-likelihood graph (ablation)Faster-RCNN · VG1000AP: 8.2 -> 8.5Improved
add context-likelihood graph (ablation)DETR · VG1000AP: 5.8 -> 6.6Improved
add energy-based graph refinement (ablation)DETR · VG1000AP: 6.6 -> 6.8Improved
use ground truth relation set instead of predicted (oracle) (ablation)Faster-RCNN · VG1000AP: 7.3 -> 15.3Improved
increase DETR object queries from 200 to 300 (ablation)DETR with CL graph and refinement · VG1000AP: 6.6 -> 7.2Improved
increase Faster-RCNN proposals from 256 to 512 (ablation)F-RCNN with CL graph and refinement · VG1000AP: 8.0 -> 8.2No clear effect
add context-likelihood graph with refinementFaster-RCNN · VG1000fps: 14.6 -> 9.4 fpsHurt
add context-likelihood graph with refinementFaster-RCNN · MS-COCOAP: 38.3 -> 42.5Improved
add context-likelihood graph with refinementDETR · MS-COCOAP: 39.9 -> 40.6Improved
add context-likelihood graph with refinementFaster-RCNN rare classes · VG1000per-class relative improvementImproved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 0 of 1 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