Research map / Label-efficient and robust learning / Recognition and long-tail
Automated Detection of Cat Facial Landmarks
George Martvel, Ilan Shimshoni, Anna Zamansky
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 changed | Where | Measured | Effect |
|---|---|---|---|
| use detected face boxes (EfficientNetV2B0 detector) instead of pre-cropped faces (ablation) | ELD · CatFLW test | NME: 2.98 -> 5.09 | Hurt |
| use YOLOv8n face detector instead of EfficientNetV2 detector (ablation) | ELD · CatFLW test | NME: 5.09 -> 4.08 | Improved |
| use eye region size 1/4 instead of 1/2 (ablation) | ELD · CatFLW test | NME: 3.11 -> 2.98 | Improved |
| use eye region size 1/4 instead of 1/6 (ablation) | ELD · CatFLW test | NME: 3.08 -> 2.98 | Improved |
| use nose region size 1/2 instead of 1 (ablation) | ELD · CatFLW test | NME: 3.31 -> 2.98 | Improved |
| use ear region size 1/2 instead of 1/3 (ablation) | ELD · CatFLW test | NME: 3.21 -> 2.98 | Improved |
| use EfficientNetV2S backbone instead of MobileNetV2 (ablation) | ELD · CatFLW test | NME: 3.09 -> 2.83 | Improved |
| use ConvNeXtTiny backbone instead of EfficientNetV2B0 (ablation) | ELD · CatFLW test | NME: 2.98 -> 2.78 | Improved |
| use ResNet101 backbone instead of ConvNeXtTiny (ablation) | ELD · CatFLW test | NME: 2.78 -> 2.91 | Hurt |
| use ensemble ELD instead of Stacked Hourglass | Stacked Hourglass · CatFLW test | NME: 3.78 -> 2.96 | Improved |
| use ensemble ELD instead of DLC ResNet101 | DLC · CatFLW test | NME: 3.81 -> 2.96 | Improved |
| apply ELD to human faces with all landmarks instead of subset (ablation) | ELD · WFLW | NME: 4.05 -> 4.65 | Hurt |
Closest papers by meaning
- AnimalWeb: A Large-Scale Hierarchical Dataset of Annotated Animal Faces (CVPR 2020 · similarity 0.86)
- Facial Landmark Detection: A Literature Survey (IJCV 2018 · similarity 0.86)
- Supervised Neural Style Transfer as an Augmentation Technique for Facial Landmark Detection (IJCV 2026 · similarity 0.85)
- Facial Landmark Detection with Tweaked Convolutional Neural Networks (TPAMI 2017 · similarity 0.85)
- PetFace: A Large-Scale Dataset and Benchmark for Animal Identification (ECCV 2024 · similarity 0.84)
- Brute-Force Facial Landmark Analysis With a 140,000-Way Classifier (AAAI 2018 · similarity 0.84)
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.
