AI Paper Atlas

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

Exploring Why Object Recognition Performance Degrades Across Income Levels and Geographies with Factor Annotations

Laura Gustafson, Megan Richards, Melissa Hall, Caner Hazirbas, Diane Bouchacourt, Mark Ibrahim

NeurIPS 2023 · Detection and segmentation › Open-set and few-shot detection · approach: Object-level representation learning · 0 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 texture-bias mitigation (Geirhos et al. training) to ResNet-50ResNet-50 ImageNet-1k · Dollar Street overalltop-5 accuracy: 32.2 -> 33.0Improved
add texture-bias mitigation to ResNet-50ResNet-50 ImageNet-1k · Dollar Street texture-marked images, low incometop-5 accuracy: 25.2 -> 29.3Improved
add texture-bias mitigation to ResNet-50ResNet-50 ImageNet-1k · Dollar Street texture-marked images, Africatop-5 accuracy: 18.6 -> 21.6Improved
add texture-bias mitigation to ResNet-50ResNet-50 ImageNet-1k · Dollar Street texture-marked images, Asiatop-5 accuracy: 24.4 -> 25.0Improved
vary the income level of Dollar Street images (high vs low)CLIP ViT-B/32 · Dollar Streettop-5 accuracy: 92.6 -> 66.9Hurt
switch CLIP image encoder from ViT-B/32 to ResNet50CLIP · Dollar Street low incometop-5 accuracy: 66.9 -> 73.4Improved
switch CLIP image encoder from ViT-B/32 to ResNet101CLIP · Dollar Street low incometop-5 accuracy: 66.9 -> 68.2Improved
use large-scale self-supervised pretraining (SEER) instead of supervised ResNet50 IN21kimage classifier · Dollar Street low incometop-5 accuracy: 58.4 -> 64.3Improved
replace ViT with MLP Mixer (both IN21k)image classifier · Dollar Street low incometop-5 accuracy: 61.3 -> 61.9No clear effect
replace supervised ViT/ResNet by BEiTimage classifier · Dollar Street low incometop-5 accuracy: 61.3 -> 51.7Hurt
higher image resolution (DPI) as a predictor of accuracyCLIP · Dollar StreetPearson correlation with top-5 accuracy: -0.035 (p = 0.00005)No clear effect
larger ImageNet-21K class count as a predictor of accuracyBEiT · Dollar StreetPearson correlation with top-5 accuracy: 0.222 (p < 0.0001)No clear effect

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