AI Paper Atlas

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

WeDetect: Fast Open-Vocabulary Object Detection as Retrieval

Shenghao Fu, Yukun Su, Fengyun Rao, Jing LYU, Xiaohua Xie, Wei-Shi Zheng

CVPR 2026 · Detection and segmentation › Open-set and few-shot detection · approach: Open-vocabulary object detection · 10 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 coarse-grained labels to multi-granularity training (ablation)WeDetect-Base · LVIS minivalAP: 46.4 -> 47.3Improved
add fine-grained labels to multi-granularity training (ablation)WeDetect-Base · LVIS minivalAP: 45.1 -> 47.3Improved
use staged-wise training (initialize neck/head first) (ablation)WeDetect-Base · LVIS minivalAP: 45.5 -> 47.3Improved
use sigmoid focal loss instead of BCE (ablation)WeDetect-Ref 4B · COCOAP: 43.1 -> 47.5Improved
add negative detection data (ablation)WeDetect-Ref 4B · COCOAP: 42.1 -> 47.5Improved
increase tokens per object from 1 to 25 (ablation)WeDetect-Ref 4B · COCOAP: 47.5 -> 47.6No clear effect
shuffle proposal order (ablation)WeDetect-Ref 4B · COCOAP: 47.5 -> 47.6No clear effect
replace fusion-based detectors with non-fusion dual-tower WeDetect-TinyWeDetect-Tiny vs YOLO-World-L · LVIS minivalAP: 35.4 -> 37.4Improved
use LLM as one-pass classifier over proposals instead of next-token predictionWeDetect-Ref 4B vs Qwen3-VL 4B · RefCOCO/+/g averageaccuracy: 86.6 -> 93.2Improved
use object-level cached embeddings (WeDetect-Uni) vs image-level CLIP-style retrievalWeDetect-Large-Uni vs FG-CLIP2 · COCO object retrievalF1: 57.7 -> 83.6Improved
use WeDetect-Large-Uni vs UPN on LVIS at 100 proposalsWeDetect-Large-Uni · LVISAR50: 65.0 -> 70.9Improved
use WeDetect-Large-Uni vs UPN on COCO AR50 at 100 proposalsWeDetect-Large-Uni · COCOAR50: 90.6 -> 89.7Hurt

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