Face Recognition – Create prototype and high level technical documentation - App project

Mobile app project description

Face Recognition – Create prototype and high level technical documentation

We need a prototype for an independent self-developed face recognition which must be high scalable, fast, secured with high precision. The business case is to search for an optical twin by uploading a picture of the user in a database of images from the user community (more details will be shared later).

Actual used technology yet:
Microsoft Azure Face Recognition
AWS Cloud as backend
Problems of the current solution:
- Each hits against MS Face Recognition costs money
- Too slow with max. 200 hits per second
- Accuracy not high enough
- Not comfortable configurable
- No own further development possible
This solution now is used in a app for iOS and Android which is available since over 1 year. Currently there are over 10.000 registered users.
To be independent from third provider solution, the goal is to have a own self developed face recognition.
The new face recognition algorithm should have following main functions:

- Face detection (in larger images scan for face)
- Face verification (male, female, identify real person, avoid cartoons, non-humans persons)
- Face identification (by all parts of faces like: eye, eyebrow, cheek, nose, mouth, chin, jaw, forehead, head etc.)
- Face Grouping (Allowing grouping face by face parts (like all with the same eyes or nose etc.)
- Similar Face searching by setting up levels in percentages (this allow later to show user faces with minimum face matching or maximum matching)
- Above the listed function, it should have at least minimum the same function like Microsoft Face Recognition.
The face itself should be check/recognize all parts of the face and filterable for each of a face part.

The face recognition must be fast, secured and compatible for multi platforms (web, iOS and Android). It must work for Millions of photos and should be high scalable and secured.

We need a prototype of the solution with high level estimations of the cost for the final developing and also the timeline.

This face recognition must than fully integrated in the already existing environment and fully work with the app which are available on iOS and Android (details will be later shared)
Also a redesign of the app is needed to be conform with the market standard and compatible with the latest devices. For this case we would like to see also a some design recommendation and recommended placing of the functionalities and improvements in usability etc for the existing app.
Please be aware, that generic messages/offers will be not answered. We are searching for expertise in ai development and app/website development.

This project is only for creating a prototype and detailed proper document how the improved solution could be (with tests and results, estimations of speed for the searches cases, UI, backend, architecture, design of app, security approach, scalability).
How the process for this project is planned:
1. Start biding on creating the proposal and prototype
2. 2 or 3 best matching company will get the first initial project for creating the complete technical solution and prototype (details listed before), timeline, cost etc.
3. One company with the best solution will be selected to create this solution finally.
The total budget for creating the final proper technical proposal with a prototype and redesign recommendation for the app will be defined later.
The total final project budget depends, the focus is keeping the scope and budget itself. Please give us a rough estimations for it.

Final starting time of this project is at the moment planed for January 2020. The company is located in Germany.
Please be aware of the obligated NDA and overhanding the whole project at the end of the project, where the copyrights etc. and the whole output will be overhanded to the owner of this project.