Research map / Image generation / Autoregressive and latent generation
Image Cropping with Composition and Saliency Aware Aesthetic Score Map
Yi Tu, Li Niu, Weijie Zhao, Dawei Cheng, Liqing Zhang
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 |
|---|---|---|---|
| use ASM-Net with composition- and saliency-aware score maps | VGG16-based image cropping model · FCDB | IoU: 0.7349 -> 0.7489 | Improved |
| use ASM-Net with composition- and saliency-aware score maps | VGG16-based image cropping model · FLMS | IoU: 0.8365 -> 0.8486 | Improved |
| use ASM-Net with composition- and saliency-aware score maps | VGG16-based image cropping model · GAICD | SRCC: 0.735 -> 0.766 | Improved |
| add composition-aware partitioning instead of using one shared score per region (ablation) | ASM-Net on CPC · CPC validation | Accuracy: 59.46 -> 87.59 | Improved |
| use horizontal-aware 1×3 grid rather than vertical-aware 3×1 grid (ablation) | ASM-Net on CPC · CPC validation | Accuracy: 71.33 -> 75.64 | Improved |
| use cross-rectangle composition pattern rather than 5×5 grid (ablation) | ASM-Net on CPC · CPC validation | Accuracy: 86.38 -> 87.59 | Improved |
| use cross-rectangle rather than multi-rectangle composition pattern (ablation) | ASM-Net on CPC · CPC validation | Accuracy: 81.34 -> 87.59 | Improved |
| use cross-rectangle rather than multi-direction composition pattern (ablation) | ASM-Net on CPC · CPC validation | Accuracy: 83.22 -> 87.59 | Improved |
| add saliency-aware loss with λ=0.1 (ablation) | ASM-Net on CPC · CPC validation | Accuracy: 3.22% | Improved |
| increase saliency-loss trade-off λ beyond 0.1 (ablation) | ASM-Net on CPC · CPC validation | Accuracy | Hurt |
Closest papers by meaning
- A Deep Network Solution for Attention and Aesthetics Aware Photo Cropping (TPAMI 2018 · similarity 0.93)
- Deep Cropping via Attention Box Prediction and Aesthetics Assessment (ICCV 2017 · similarity 0.93)
- Venus: Benchmarking and Empowering Multimodal Large Language Models for Aesthetic Guidance and Cropping (CVPR 2026 · similarity 0.87)
- Spatial-Semantic Collaborative Cropping for User Generated Content (AAAI 2024 · similarity 0.87)
- Cropper: Vision-Language Model for Image Cropping through In-Context Learning (CVPR 2025 · similarity 0.87)
- Learning Subject-Aware Cropping by Outpainting Professional Photos (AAAI 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.
