Research map / Detection and segmentation / Open-set and few-shot detection
Unsupervised Part-Based Disentangling of Object Shape and Appearance
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 unsupervised part-based shape and appearance learning instead of the strongest reported landmark baseline | Unsupervised landmark prediction · Cat Head, 10 landmarks | landmark error (% inter-ocular distance): 15.35 -> 9.88 | Helped |
| Use unsupervised part-based shape and appearance learning instead of the strongest reported landmark baseline | Unsupervised landmark prediction · Cat Head, 20 landmarks | landmark error (% inter-ocular distance): 14.84 -> 9.30 | Helped |
| Use unsupervised part-based shape and appearance learning instead of the strongest reported landmark baseline | Unsupervised landmark prediction · MAFL, 10 landmarks | landmark error (% inter-ocular distance): 3.19 -> 3.24 | Hurt |
| Use unsupervised part-based shape and appearance learning instead of the strongest reported landmark baseline | Unsupervised landmark prediction · CUB-200-2011, 10 landmarks | landmark error (% image edge length): 5.36 -> 3.91 | Helped |
| Use unsupervised part-based shape and appearance learning instead of the strongest reported unsupervised baseline | Unsupervised landmark prediction · BBC Pose test | PCK at 6 pixels: 68.4\% -> 74.5\% | Helped |
| Use unsupervised part-based shape and appearance learning instead of the strongest reported unsupervised baseline | Unsupervised landmark prediction · Human3.6M test, 16 landmarks | landmark error (% image size): 4.91 -> 2.79 | Helped |
| Use unsupervised part-based synthesis instead of supervised VU-Net | Shape/appearance-swapped image generation · Deep Fashion test | mAP: 88.7\% -> {90.3}\% | Helped |
| Use unsupervised part-based synthesis instead of supervised VU-Net | Shape/appearance-swapped image generation · Deep Fashion test | rank-1 accuracy: 87.5\% -> {89.4}\% | Helped |
| Use unsupervised part-based synthesis instead of supervised VU-Net | Shape/appearance-swapped image generation · Deep Fashion test | rank-5 accuracy: {98.7}\% -> {98.2}\% | Hurt |
| Use unsupervised part-based synthesis instead of supervised VU-Net | Shape/appearance-swapped image generation · Deep Fashion test | rank-10 accuracy: {99.5}\% -> {99.2}\% | Hurt |
| Use unsupervised part-based synthesis instead of supervised VU-Net | Pose transfer · Deep Fashion test, α=2.5% | PCK: {95.2}\% -> 85.6\% | Hurt |
| Use unsupervised part-based synthesis instead of supervised VU-Net | Pose transfer · Deep Fashion test, α=5% | PCK: {98.4}\% -> 94.2\% | Hurt |
Closest papers by meaning
- Unsupervised Part Segmentation Through Disentangling Appearance and Shape (CVPR 2021 · similarity 0.92)
- Generating Furry Cars: Disentangling Object Shape and Appearance across Multiple Domains (ICLR 2021 · similarity 0.89)
- Unsupervised Part Discovery from Contrastive Reconstruction (NeurIPS 2021 · similarity 0.89)
- Articulate your NeRF: Unsupervised articulated object modeling via conditional view synthesis (NeurIPS 2024 · similarity 0.87)
- Learning Partonomic 3D Reconstruction from Image Collections (CVPR 2025 · similarity 0.87)
- Visual Anagrams Reveal Hidden Differences in Holistic Shape Processing Across Vision Models (NeurIPS 2025 · similarity 0.86)
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.