Brief: Build a controllable synthetic degradation pipeline (sand/dust turbulence, heat shimmer, rain, motion blur) and a curated benchmark dataset and evaluation protocol to support a drone‑based computer‑vision project for detection, restoration, and feature extraction in adverse conditions.

Goals and responsibilities:

• Implement physically motivated degradation generators with tunable severity.
• Curate a benchmark dataset with clean/degraded variants, consistent labeling, and leakage‑free train/val/test splits.
• Define the evaluation protocol and utilities (metrics, scoring scripts, reporting) for image quality and downstream tasks; produce baseline stats and visualizations.
• Deliver a clean, reusable package with thorough documentation.

Required skills:

• Strong Python; image/video processing libraries; CV basics (image formation, blur kernels, noise models).
• Engineering best practices: reproducibility, configuration management, dataset versioning. Nice‑to‑have: image restoration/deblurring, atmospheric optics/turbulence, annotation tools.

Planned training:

• Direct supervision by a PFE student; onboarding to project data conventions and evaluation needs; guided readings on turbulence/atmospheric degradation and aerial CV benchmarks.

Other details:

• Targeting students entering their final year and interested in a PFE with ReDX.
• Recommended period: 3 months.
• Compensation: Monthly stipend.