Research map / Label-efficient and robust learning / Recognition and long-tail
Rethinking Image Cropping: Exploring Diverse Compositions From Global Views
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 |
|---|---|---|---|
| predict multiple globally regressed crops with learnable anchors and validity classification | Set prediction image cropping model · GAICv1 | ACC5: 66.5 -> 81.5 | Helped |
| predict multiple globally regressed crops with learnable anchors and validity classification | Set prediction image cropping model · GAICv2 | ACC5: 69.0 -> 85.0 | Helped |
| predict multiple globally regressed crops with learnable anchors and validity classification | Set prediction image cropping model · GAICv2 | AP (K = 40): 42.2 -> 56.8 | Helped |
| predict multiple globally regressed crops with learnable anchors and validity classification | Set prediction image cropping model · FLMS | IoU: 0.837 -> 0.838 | No clear effect |
| add self-distillation label smoothing (ablation) | Set prediction image cropping model · GAICv2 | AP (K = 40): 54.8 -> 55.5 | Helped |
| add self-distillation label smoothing (ablation) | Set prediction image cropping model · CPC | AP (K = 40): 26.6 -> 27.8 | Helped |
| use quality-guided smoothing rather than self-distillation smoothing (ablation) | Set prediction image cropping model · GAICv2 | AP (ϵ = 0.90, K = 40): 43.5 -> 47.4 | Helped |
| apply quality-guided smoothing rather than hard binary labels (ablation) | Set prediction image cropping model · GAICv2 | AP (ϵ = 0.90, K = 40): 42.8 -> 47.4 | Helped |
| raise the quality-guided smoothing upper bound to µ = 0.9 (ablation) | Quality-guided label smoothing · GAICv2 | AP (ϵ = 0.90, K = 40): 29.2 | Hurt |
Closest papers by meaning
- Image Cropping With Spatial-Aware Feature and Rank Consistency (CVPR 2023 · similarity 0.89)
- Reliable and Efficient Image Cropping: A Grid Anchor Based Approach (CVPR 2019 · similarity 0.89)
- Cropper: Vision-Language Model for Image Cropping through In-Context Learning (CVPR 2025 · similarity 0.88)
- Learning Subject-Aware Cropping by Outpainting Professional Photos (AAAI 2024 · similarity 0.87)
- Spatial-Semantic Collaborative Cropping for User Generated Content (AAAI 2024 · similarity 0.85)
- Combining Deep Learning and Street View Imagery to Map Smallholder Crop Types (AAAI 2024 · similarity 0.85)
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.