AI Paper Atlas

Research map / Label-efficient and robust learning / Recognition and long-tail

AnimalWeb: A Large-Scale Hierarchical Dataset of Annotated Animal Faces

Muhammad Haris Khan, John McDonagh, Salman Khan, Muhammad Shahabuddin, Aditya Arora, Fahad Shahbaz Khan, Ling Shao, Georgios Tzimiropoulos

CVPR 2020 · Label-efficient and robust learning › Recognition and long-tail · approach: Margin-based face recognition losses · 52 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
Evaluate alignment on unseen species rather than species represented in trainingHourglass on AnimalWeb · AnimalWeb, 9-point alignmentNME (%): 5.23 -> 6.44Hurt
Use HG-3 instead of HG-2 (ablation)Hourglass on AnimalWeb · AnimalWeb Known species, 9-point alignmentNME (%): 5.35 -> 5.23Improved
Use HG-3 instead of HG-2 (ablation)Hourglass on AnimalWeb · AnimalWeb Unknown species, 9-point alignmentNME (%): 6.50 -> 6.44Improved
Increase ResNet depth across the tested variants (ablation)Fine-grained species recognition on AnimalWeb · AnimalWebtop-1 accuracy (%): 76.49 -> 79.22 -> 80.04 -> 81.06Improved
Evaluate at extreme yaw angles rather than the central [-45°,45°] rangeHG-3 on AnimalWeb Known species · AnimalWeb face alignment by yawNME (%): 3.31; 7.35 at -90°; 6.96 at 90°Hurt
Use the 9-point rather than the 68-point landmark setup on human alignment benchmarks (ablation)Hourglass · 300W challengingNME (%): 2.25 (9 pts., HG-3); 2.01 (68 pts., HG-3)Mixed
Use the 9-point rather than the 68-point landmark setup (ablation)Hourglass · COFWNME (%): 3.26 (9 pts., HG-3); 2.60 (68 pts., HG-3)Hurt

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