Proof of Play (PoP) Super-Resolution & Audit Engine

Delivered scope

• Features billboard ROI auto-detection and custom polygon bounding box parsing.
• Transfers EXIF metadata and computes SHA-256 cryptographic hashes for audit-compliant verification.
• Includes FastAPI web interface, CLI tooling, memory governor, and comprehensive pytest test suites.

Technology

System approach

• API & Core: FastAPI service with modular CLI, processing pipelines, and Dockerized deployment.
• Vision Pipeline: OpenCV-based billboard ROI detection and adaptive contrast, denoising, and super-resolution models.
• Integrity Layer: SHA-256 file hashing before and after enhancement, combined with lossless EXIF metadata cloning.
• Memory Governor: Dynamic RAM and VRAM throttling to ensure high-throughput batch execution without worker crashes.

Challenges

• Enhancing nighttime and distant billboard photos without introducing visual artifacts that could invalidate compliance audits.
• Preserving legal integrity of proof photos when metadata is typically lost during neural network image processing.
• Managing memory spikes when processing batches of 24MP+ high-resolution camera proofs concurrently.

Solutions

• Implemented ROI-focused enhancement that isolates the billboard face before applying adaptive contrast and sharpening filters.
• Engineered an EXIF metadata cloner that transfers exact camera, timestamp, and GPS headers onto the enhanced output.
• Built a proactive Memory Governor that scales batch concurrency according to available system memory and image resolution.

Key outcomes