Research map / Vision-language and multimodal / CLIP and zero-shot
Modeling the Visual Ambiguity of Human Sketches
Yang Zhou, Ping Ni, Jin Wang, Senyun Jia, Jingdan Yan, Kaixiang Huang, Guodong Lu, Jingru Yang, Shengfeng He
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 Elastic Matching (AmbiScore-based soft margin) (ablation) | DisAmb · Sketchy | mAP@200: 0.514 -> 0.682 | Improved |
| add Elastic Matching (ablation) | DisAmb · Sketchy ZS-FG-SBIR | acc@1: 17.24 -> 28.09 | Improved |
| add texture swapping (ablation) | DisAmb · Sketchy ZS-FG-SBIR | acc@1: 21.28 -> 28.09 | Improved |
| add shape jigsaw loss (L_jigsaw) (ablation) | DisAmb · Sketchy | mAP@200: 0.616 -> 0.682 | Improved |
| add shape jigsaw pair loss (L_pair) (ablation) | DisAmb · Sketchy | mAP@200: 0.631 -> 0.682 | Improved |
| add extra 'person' prompt in AmbiScore (ablation) | DisAmb · Sketchy | mAP@200: 0.651 -> 0.682 | Improved |
| vary jigsaw segments k from 16 (ablation) | DisAmb · Sketchy | mAP@200: k=16 0.682; k=4 0.603; k=36 0.580 | Hurt |
| add texture swapping to baseline trained on images (ablation) | ViT-B/16 baseline · Sketchy | mAP@200: 0.479 -> 0.537 | Improved |
| remove texture (train on masks or edge maps instead of images) (ablation) | ViT-B/16 baseline · Sketchy | mAP@200: 0.479 -> 0.466 (mask), 0.449 (edge) | No clear effect |
| use DisAmb SBIR model as supervision for segmentation | SketchYourSeg · Sketchy | pAcc: 68.29 -> 79.86 | Improved |
| use DisAmb SBIR model as supervision for generation | StableSketch · Sketchy | FID-I: 26.21 -> 21.05 | Improved |
| train on highly ambiguous pairs | existing SBIR methods · Sketchy Ext | mAP | Hurt |
Closest papers by meaning
- SEVA: Leveraging sketches to evaluate alignment between human and machine visual abstraction (NeurIPS 2023 · similarity 0.90)
- SketchAA: Abstract Representation for Abstract Sketches (ICCV 2021 · similarity 0.89)
- Sketch Down the FLOPs: Towards Efficient Networks for Human Sketch (CVPR 2025 · similarity 0.87)
- Open Vocabulary Semantic Scene Sketch Understanding (CVPR 2024 · similarity 0.87)
- SketchFusion: Learning Universal Sketch Features through Fusing Foundation Models (CVPR 2025 · similarity 0.87)
- It's All About Your Sketch: Democratising Sketch Control in Diffusion Models (CVPR 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.
