AI Paper Atlas

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

SRA-Det: Learning Omni-Grained Open-Vocabulary Detection Beyond Category Names

Li Yang, Boyu Cai, Wei Liu, Yan Wang, Chunfeng Yuan, Bing Li, Weiming Hu

CVPR 2026 · Detection and segmentation › Open-set and few-shot detection · approach: Open-vocabulary object detection

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 Semantic Retrieval Augmentation (multi-facet text + soft-min matching) (ablation)SRA-Det · FG-OVD HardmAP: 40.8 -> 43.1Improved
add Semantic Retrieval Augmentation (ablation)SRA-Det · LVIS minivalAP: 28.5 -> 28.5No clear effect
add attribute-augmented data (LLM + CLIP dual check) to V3Det (ablation)SRA-Det · FG-OVD HardmAP: 25.3 -> 30.3Improved
add attribute-augmented data (ablation)SRA-Det · LVIS minivalAP: 27.6 -> 27.9No clear effect
add VAW attribute dataset (ablation)SRA-Det · FG-OVD HardmAP: 30.3 -> 43.1Improved
add Objects365 and GoldG (ablation)SRA-Det · LVIS minivalAP: 28.5 -> 40.4Improved
add Objects365 and GoldG (ablation)SRA-Det · FG-OVD HardmAP: 43.1 -> 45.2Improved
initialize retrieval queries with learnable query plus previous retrieval result instead of learnable query only (ablation)SRA module · FG-OVD HardmAP: 42.3 -> 43.1Improved
initialize retrieval queries with learnable query plus previous retrieval result instead of MLP(EOS) (ablation)SRA module · FG-OVD HardmAP: 42.6 -> 43.1Improved
combine learnable query with EOS embedding instead of previous retrieval result (ablation)SRA module · FG-OVD HardmAP: 43.0 vs 43.1No clear effect
fine-tune on FG-OVD training setSRA-Det · FG-OVD averagemAP: 54.9 -> 67.8Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 4 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