Tuned MRZ reader source code for Android

I need an Android project with source code which is able to detect MRZ lines efficiently on video stream, using the native camera driver. If an MRZ found on a frame, it parses its alphanumeric data, calculates and compares all check digits and displays all fields. Requirements (in roomlight conditions) : - reads one line (driving license), two line (passport), three line (ID card) of zone - capture&detection in 5 seconds - min 80% success rate (at least 8 successful detection out of 10 tries) - open-source libs only, restrictive licence is not acceptable - optionally: auto-correction of common character mistakes (8/B, 2/Z, 0/O, 4/H, 1/I ) until check digit validity passes (you may use my GitHub project ZsBT/mrz-java ) I do NOT need the facial image, this job is about the alphanumeric data only. Do not expect special input: the card image samples I have strictly meet the ICAO standards (OCRB, 1/2/3 lines, field structure, etc). Maximum image tilt/yaw will not exceed 10%. Efficiency is important, the official JMRTD library I used so far is too slow and unreliable. Developers who already have a working solution (e.g. it was one of his/her previous project) are preferred. In this case, please show me a demo app (I declare I won't reverse-engineer that) or a proof of videoclip. Upon agreement, I expect the above-mentioned deliverable in two weeks. I need you to waive the owner/author rights of the source code. Feel free to use existing open-source project (JMRTD, tesseract libs), and fine-tune for convenient use. You do not need to make it from scratch, therefore the planned budget of this project is only 300 USD. Candidates who have no experience will be ignored. Please include the text "I indeed read the description" when you apply. General cover letters are ignored, I would like to hear your planned tools, experience and thoughts about this specific project.