AI Paper Atlas

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

Learning To Predict Visual Attributes in the Wild

Khoi Pham, Kushal Kafle, Zhe Lin, Zhihong Ding, Scott Cohen, Quan Tran, Abhinav Shrivastava

CVPR 2021 · Detection and segmentation › Open-set and few-shot detection · approach: Object-level representation learning · 90 citations (OpenAlex)

Check 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 negative label expansion (ablation)Strong Baseline · VAW testmAP: 65.9 -> 67.7Improved
Add supervised contrastive learning after negative label expansion (ablation)Strong Baseline · VAW testmAP: 67.7 -> 68.2Improved
Add reweighting and repeat factor sampling after negative expansion and SupCon (ablation)Strong Baseline · VAW testmR@15: 55.2 -> 58.3Improved
Apply SCoNE (ablation)ResNet-Baseline · VAW testmAP: 63.0 -> 66.4Improved
Add multi-attention (ablation)Strong Baseline · VAW testmAP: 67.4 -> 67.7Improved
Add low-level features (ablation)Strong Baseline · VAW testmAP: 67.3 -> 67.7Improved
Add the object localizer (ablation)Strong Baseline · VAW testmAP: 66.9 -> 67.7Improved
Add object localizer, multi-attention, and low-level features (ablation)ResNet-Baseline · VAW testmAP: 65.6 -> 67.7Improved
Use SupCon pretraining instead of joint SupCon training (ablation)Strong Baseline with multi-attention · VAW testmAP: 67.7 -> 67.3Hurt
Jointly train with SupCon loss (ablation)Strong Baseline with multi-attention · VAW testmAP: 67.7 -> 68.2Improved
Use SupCon pretraining without multi-attention (ablation)Strong Baseline without multi-attention · VAW testmAP: 67.4 -> 67.6Improved
Use class-aware sampling (ablation)ResNet-Baseline after negative label expansion · VAW testmAP: 65.6 -> 63.5Hurt

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