Image Fraud Detection -- 2

We have an existing ID card validation system which captures the ID book and compares it to a selfie taken of the user. We are implementing various image fraud detection algos. Our goal is to have an automated image forensics test based on existing Python or C++ libraries that does not use machine learning. We want to use classical approaches with no machine learning. You need to implement three apps: 1. "Error Level Analysis" approach developed by Dr Niel Krawetz. 2. Metadata test - extract metadata (EXIF etc) and confirm that the photo isn't from photoshop, paint etc.(can use [login to view URL]) 3. Do a test for straight lines around the identity number. In our minds, this would be to run a Canny Edge detector and then do a Hough Transform to pick up the straight lines Experience in ELA is fundamental - do not apply if you don't have previous ELA experience. You will need to sign an non-disclosure & copyright transfer agreement. We expect the work to be 2 days. We are looking for an Eastern European developer