BOLT Vision Gate

BOLT Vision Gate is an OpenCV 5 proof of concept that analyzes brightness, contrast, blur variance and edge density before a visual input is allowed into an automated workflow. Poor or ambiguous frames are escalated to human review with explicit machine-readable reasons. The project is implemented in Python with deterministic samples, automated tests, a technical report and reproducible packaging. It is now deployed on AWS Lambda in eu-north-1 with a live public endpoint. A live validation returned OpenCV 5.0.0 and a structured human_review decision from the AWS-hosted service. The design is intentionally bounded: it does not claim general visual safety or perform biometric identity inference. The main implementation challenges were adapting to OpenCV 5, packaging native computer-vision dependencies for AWS Lambda arm64, and keeping the safety gate transparent and reproducible.

Built With

• aws-lambda
• computer-vision
• docker
• human-in-the-loop
• opencv-5
• python
• responsible-ai