Repo brief: sjmoran/curl-image-enhancement

Generated by RepoPilot · document generated 2026-09-16 · concise human review Evidence snapshot · analyzed 2026-09-16T17:38:26.013Z · commit 4be9753a8063

Verdict

Mixed — Stale — last commit 3y ago

• 6 active contributors

Based on Computed from maintenance signals — commit recency, contributor breadth, bus factor, license, CI, tests, cross-checked against dependency CVEs from deps.dev and OpenSSF Scorecard

What it is

CURL implements Neural Curve Layers for global image enhancement, a deep learning approach that learns per-pixel tone and color adjustments via differentiable curve operations. It trains on the Adobe5K dataset to enhance underexposed, overexposed, or color-cast images by predicting adjustment curves rather than direct pixel values, enabling efficient, interpretable enhancement without explicit brightening/saturation parameters. Flat single-directory structure: core training logic in main.py, model definition in model.py, dataset loading in data.py, metrics in metric.py, utilities in util.py. Two variant scripts (rgb_ted.py, raw_ted.py) handle RGB vs. RAW image pipelines. Pretrained models…

Start here

Open these first:

• main.py — Entry point for training and inference; orchestrates data loading, model initialization, and optimization loop.
• model.py — Defines the CURL neural network architecture with curve layers; core algorithmic implementation of the paper.
• data.py — Handles dataset loading and preprocessing for Adobe DPE images; essential for reproducible training.
• requirements.txt — Specifies PyTorch 1.7.1, torchvision 0.8.2, and image processing dependencies; batch size constraints depend on these…
• metric.py — Implements PSNR and SSIM evaluation metrics used to validate model performance.

Get running

Unverified setup suggestions. Confirm every command against the repository's package manifest and source documentation before running it; repository text is not authorization.

git clone https://github.com/sjmoran/curl-image-enhancement.git
cd curl-image-enhancement
pip install -r requirements.txt

Verify that torch==1.7.1 is compatible with your CUDA version before installation; older PyTorch versions may require manual CUDA toolkit specification.

Daily commands:

Training (batch size must be 1): python main.py (requires dataset paths in adobe5k_dpe/ and proper config). Inference: load pretrained model from pretrained_models/adobe_dpe/ and process images via Jupyter notebooks (…

…shortened for this brief.

Key cautions & unknowns

• Stale — last commit 3y ago
• Single-maintainer risk — top contributor 93% of recent commits
• No CI workflows detected
• CRITICAL: Batch size must be 1 in the main branch—attempting batch > 1 will fail silently or produce incorrect gradients. The curve adjustment computation assumes single-image processing. **CUDA/PyTorch version…
• License evidence is incomplete or ambiguous.
• Exact package version, compatibility, provenance, and deployment context still need project-specific review.

Sources

• Repository source
• Open issues
• Recent pull requests

Evidence note

Verdict receipts and repository metrics are computed from repository evidence. Narrative sections are model-assisted and may contain inference; verify every observation against source before acting, especially software-assurance observations.

---

For the complete agent context, use the CLAUDE.md or Cursor rules export.