Research map / Vision-language and multimodal / Human and embodied modeling
FS-COCO: Towards Understanding of Freehand Sketches of Common Objects in Context
Pinaki Nath Chowdhury, Aneeshan Sain, Ayan Kumar Bhunia, Tao Xiang, Yulia Gryaditskaya, Yi-Zhe Song
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 |
|---|---|---|---|
| train on FS-COCO instead of SketchyCOCO | Siam.-VGG16 · FG-SBIR tested on FS-COCO | R@1: <0.1 -> 23.3 | Improved |
| train on FS-COCO instead of SketchyScene | Siam.-VGG16 · FG-SBIR tested on FS-COCO | R@1: 1.8 -> 23.3 | Improved |
| add sketch to image-caption query (CNN-RNN -add) | CNN-RNN · FS-COCO retrieval | R@1: 11.1 -> 25.3 | Improved |
| add image caption to sketch query (CLIP* -add) vs sketch only Siam.-VGG16 | Siam.-VGG16 · FS-COCO retrieval | R@1: 23.3 -> 23.9 | Mixed |
| add H-Decoder pre-training (ablation) | Siam.-VGG16 · FS-COCO retrieval | R@1: 23.3 -> 24.1 | Improved |
| add H-Decoder pre-training (ablation) | CLIP* · FS-COCO retrieval | R@10: 26.5 -> 27.1 | Improved |
| add H-Decoder pre-training (ablation) | LNFMM · FS-COCO sketch captioning | CIDEr: 90.1 -> 95.3 | Improved |
| add H-Decoder pre-training (ablation) | LNFMM · FS-COCO sketch captioning | BLEU-4: 16.7 -> 17.3 | Improved |
| replace hierarchical decoder with single-layer RNN on RDP-simplified sketches (ablation) | Siam.-VGG16 · FS-COCO retrieval | R@10: 54.3 -> 52.1 | Hurt |
| mask early strokes instead of late strokes at test time | SBIR on FS-COCO · FS-COCO retrieval | R@10 | Hurt |
| use sketch captions instead of image captions as query | CLIP* · FS-COCO retrieval | R@1: 22.1 -> 14.8 | Hurt |
| use sketches instead of image captions as query (fine-tuned) | Siam.-VGG16 vs CLIP* · FS-COCO retrieval | R@1: 22.1 -> 23.3 | Improved |
Closest papers by meaning
- Open Vocabulary Semantic Scene Sketch Understanding (CVPR 2024 · similarity 0.91)
- SketchyScene: Richly-Annotated Scene Sketches (ECCV 2018 · similarity 0.90)
- SceneSketcher: Fine-Grained Image Retrieval with Scene Sketches (ECCV 2020 · similarity 0.90)
- Scene-Level Sketch-Based Image Retrieval with Minimal Pairwise Supervision (AAAI 2023 · similarity 0.89)
- Sketch Down the FLOPs: Towards Efficient Networks for Human Sketch (CVPR 2025 · similarity 0.88)
- O3SLM: Open Weight, Open Data, and Open Vocabulary Sketch-Language Model (AAAI 2026 · similarity 0.87)
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.
