Research map / Image generation / Image restoration
Learning to Structure an Image With Few Colors
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 |
|---|---|---|---|
| use ColorCNN task-centered quantization instead of traditional color quantization | AlexNet classifier · CIFAR10, 1-bit | top-1 accuracy: +37.6% absolute accuracy | Helped |
| use ColorCNN task-centered quantization instead of traditional color quantization | AlexNet classifier · CIFAR100, 1-bit | top-1 accuracy: +8.6% absolute accuracy | Helped |
| use ColorCNN task-centered quantization instead of traditional color quantization | AlexNet classifier · STL10, 1-bit | top-1 accuracy: +9.6% absolute accuracy | Helped |
| use ColorCNN task-centered quantization instead of traditional color quantization | AlexNet classifier · Tiny200, 1-bit | top-1 accuracy: +14.5% absolute accuracy | Helped |
| add regularization that encourages all palette colors to be selected (ablation) | ColorCNN · STL10, 3-bit | accuracy: 67.5 -> 69.7 | Helped |
| add color jitter to the train-time quantized image (ablation) | ColorCNN · STL10, 3-bit | accuracy: 67.8 -> 69.7 | Helped |
| replace the U-Net backbone with DnCNN (ablation) | ColorCNN · classification tasks | top-1 accuracy | Hurt |
| set regularization weight too small or too high rather than 1 (ablation) | ColorCNN · classification tasks | top-1 accuracy | Hurt |
| set color jitter weight too small or too large rather than 1 (ablation) | ColorCNN · classification tasks | top-1 accuracy | Hurt |
| apply dithering to traditional color quantization | traditional color quantization · low-bit quantization | top-1 accuracy | No clear effect |
| use ColorCNN rather than clustering-based traditional methods in a large color space | color quantization · large color space | top-1 accuracy | Hurt |
Closest papers by meaning
- Dataset Color Quantization: A Training-Oriented Framework for Dataset-Level Compression (ICLR 2026 · similarity 0.89)
- Instance-Aware Image Colorization (CVPR 2020 · similarity 0.86)
- Learning a Deep Convolutional Network for Colorization in Monochrome-Color Dual-Lens System (AAAI 2019 · similarity 0.86)
- Color-Oriented Redundancy Reduction in Dataset Distillation (NeurIPS 2024 · similarity 0.85)
- Efficient Neural Network Encoding for 3D Color Lookup Tables (AAAI 2025 · similarity 0.84)
- ColorConceptBench: A Benchmark for Probabilistic Color-Concept Understanding in Text-to-Image Models (NeurIPS 2026 · similarity 0.84)
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.