Research map / 3D vision / Human pose estimation
General Object Pose Transformation Network from Unpaired Data
Yukun Su, Guosheng Lin, Ruizhou Sun, Qingyao Wu
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 |
|---|---|---|---|
| add perceptual loss Lperc (ablation) | pose transformation network · Mammals | mFID: 68.4 -> 33.9 | Improved |
| add content loss Lcont (ablation) | pose transformation network · Mammals | mFID: 58.5 -> 33.9 | Improved |
| add self-reconstruction loss Lself (ablation) | pose transformation network · Mammals | mFID: 36.6 -> 33.9 | Improved |
| add regularization loss Lreg (ablation) | pose transformation network · Mammals | mFID: 35.0 -> 33.9 | Improved |
| add TPS warping (ablation) | pose transformation network · Mammals | mSSIM: 0.547 -> 0.576 | Improved |
| add dense warping (ablation) | pose transformation network · Birds | mFID: 28.9 -> 28.3 | Improved |
| add style loss Lstyle (ablation) | pose transformation network · Cars | mSSIM: 0.448 -> 0.491 | Improved |
| swap SS block to decoder and TSC block to encoder positions (ablation) | pose transformation network · Human | mFID: 37.6 -> 38.2 | Hurt |
| use Lperc (ablation) | pose transformation network · Human | mSSIM: 0.255 -> 0.676 | Improved |
| augment training data with generated poses | IRNet · PASCAL VOC CAM | mIoU: 48.3 -> 49.6 | Improved |
| augment training data with generated poses | SEAM · PASCAL VOC segmentation masks (test) | mIoU: 65.7 -> 66.5 | Improved |
| our method vs best baseline on Mammals | Ours vs COCO-FUNIT · Mammals | mFID: 70.7 -> 33.9 | Improved |
Closest papers by meaning
- Pose Augmentation: Class-agnostic Object Pose Transformation for Object Recognition (ECCV 2020 · similarity 0.89)
- Synthesizing Images of Humans in Unseen Poses (CVPR 2018 · similarity 0.87)
- Towards Robust 3D Pose Transfer with Adversarial Learning (CVPR 2024 · similarity 0.87)
- NOPE: Novel Object Pose Estimation from a Single Image (CVPR 2024 · similarity 0.86)
- Source-Free and Image-Only Unsupervised Domain Adaptation for Category Level Object Pose Estimation (ICLR 2024 · similarity 0.86)
- Neural Pose Representation Learning for Generating and Transferring Non-Rigid Object Poses (NeurIPS 2024 · 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.
