AI Paper Atlas

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

FORB: A Flat Object Retrieval Benchmark for Universal Image Embedding

Pengxiang Wu, Siman Wang, Kevin S Dela Rosa, Derek Hao Hu

NeurIPS 2023 · Detection and segmentation › Open-set and few-shot detection · approach: Object-level representation learning · 1 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 reranking with local features (ablation)BoW · FORB overallmAP@5 (%): 78.44 -> 80.38Improved
Add reranking with local features (ablation)BoW · FORB overallt-mAP@5 (%): 62.49 -> 67.83Improved
Add reranking with local features (ablation)DELG · FORB overallmAP@5 (%): 48.81 -> 58.74Improved
Add reranking with local features (ablation)DELG · FORB overallinference seconds per query: 0.376 -> 6.015Hurt
Use handcrafted BoW instead of learned DELGretrieval embedding · FORB overallmAP@5 (%): 48.81 -> 78.44Improved
Use mid-level FIRe features instead of BoW low-level featuresretrieval embedding · FORB overallt-mAP@5 (%): 62.49 -> 77.50Improved
Use FIRe instead of CLIPretrieval embedding · FORB overallmAP@5 (%): 89.36 -> 88.08No clear effect
Use FIRe instead of CLIPretrieval embedding · FORB overallt-mAP@5 (%): 67.23 -> 77.50Improved
Use FIRe instead of CLIP for speedretrieval embedding · FORB inferenceseconds per query: 0.513 -> 0.124Improved
Larger model and training data for top-only methods (CLIP vs SLIP)global embedding · FORB overallmAP@5 (%): 39.01 -> 89.36Improved
Larger training data (DiHT vs DINO)global embedding · FORB overallmAP@5 (%): 55.20 -> 84.77Improved
Use top-only CLIP instead of FIRe on logos / hard splitretrieval embedding · FORB hard splitmAP@5 (%): 56.58 -> 73.84Improved

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