Face recognition
Currently I have face recognition running using the following
ageitgey/face_recognition library ([login to view URL])
Django for the API
OS Ubuntu on AWS
Every person can has only one image active
Support multiple companies
It has 2 main functionality
Verify a person: I pass person ID and base64 image, then function return found or not found
Find a person: I pass base64 image for the person, then system search in the database to find the person then return person information if found.
Other functionality is available such as adding a new person, uploading an image for a person, creating a new company, delete a person or company with all related records, list all companies and\or related, list all employees.
The problem with “Find a person” works slow due to going and searching in all table records one by one, where every time the table\database gets large, it gets slower more and more.
I need to do enhancement\rebuild for the project
Increase the speed of “Find a person” so that database records will not actually affect the speed, my actual requirements to find a person in 30,000 employees in < 1s, knowing that every person will have 5 images
Add functionality to be able to upload many pictures and train the module any time I need
Every person can has up to 5 pictures so we can increase accuracy
Support all ethnics and skin colors
Ability to save persons pictures in server hard drive or AWS S3
The project needs to deliver an enhanced Face recognition backend with REST web API using AI ML with high accuracy and speed, developed using the latest technology.
Backend (to run on Linux server)
The project need to deliver a backend (API) for face recognition using AI ML technology with high accuracy and face recognition with <1 sec for 1:N (one to many search within 30,000 person)
Source code delivery
Deliver documented source code
Deliver deployment document
Deliver the full source code
Make deployment to the production server
NOTE: NO SaaS is allowed to be use