Research map / Detection and segmentation / CNN features
DeepFlux for Skeletons in the Wild
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 |
|---|---|---|---|
| replace binary skeleton classification with context-flux regression (ablation) | DeepFlux network without ASPP · SK-LARGE | F-measure: +2.0% | Helped |
| replace binary skeleton classification with context-flux regression (ablation) | DeepFlux network without ASPP · SYM-PASCAL | F-measure: +4.9% | Helped |
| add ASPP to the binary-classification baseline (ablation) | DeepFlux network without context flux · SK-LARGE | F-measure: +1.6% | Helped |
| add ASPP to the binary-classification baseline (ablation) | DeepFlux network without context flux · SYM-PASCAL | F-measure: +1.7% | Helped |
| combine context flux and ASPP (ablation) | DeepFlux · SK-LARGE and SYM-PASCAL | F-measure: ~4% on SK-LARGE; ~10% on SYM-PASCAL | Helped |
| use context radius r=7 instead of smaller or larger tested radii (ablation) | DeepFlux · SK-LARGE and SYM-PASCAL | F-measure: SK-LARGE: 0.721, 0.727, 0.732, 0.726, 0.724; SYM-PASCAL: 0.481, 0.498, 0.502, 0.500, 0.501 | Helped |
| replace context-flux targets with binary classification on the same dilated skeleton ground truth (ablation) | DeepFlux baseline · SK-LARGE and SYM-PASCAL | F-measure: F=0.673 (-6%) on SK-LARGE; F=0.425 (-8%) on SYM-PASCAL | Hurt |
| use DeepFlux instead of Hi-Fi | skeleton detection · SK-LARGE, SK506, and WH-SYMMAX | F-measure: 0.8%, 1.4%, and 3.5% | Helped |
| use DeepFlux instead of Hi-Fi | symmetry detection · SYM-PASCAL | F-measure: 4.8% | Helped |
| use DeepFlux instead of LSN | symmetry detection · SYMMAX300 | F-measure: 1.1% | Helped |
Closest papers by meaning
- Object Skeleton Extraction in Natural Images by Fusing Scale-Associated Deep Side Outputs (CVPR 2016 · similarity 0.89)
- Object Co-Skeletonization With Co-Segmentation (CVPR 2017 · similarity 0.86)
- Multi-Scale Bidirectional FCN for Object Skeleton Extraction (AAAI 2018 · similarity 0.86)
- Real-Time Object Detection and Skeletonization for Motion Prediction in Video Streaming (AAAI 2025 · similarity 0.84)
- Learning Implicit Representation for Reconstructing Articulated Objects (ICLR 2024 · similarity 0.84)
- Stitch, Contrast, and Segment: Learning a Human Action Segmentation Model Using Trimmed Skeleton Videos (AAAI 2025 · similarity 0.83)
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.