DOCUMENTATION.md

MCQ App Face recognition

This application currently works with both password and face ID recognition.

How it works

Upon first entry into the application,

• User clicks to signup
• On signup page, user fills in information and proceeds to approve their webcamf
• Their picture is then taken with the use of both the Video and Canvas APIs provided by HTML5
• After their picture is taken, it is then encoded and coverted to a Blob string
• On the of chance that face recognition fails, the User's account is still created with a username and password.
• This data is then passed to the backend by way of HTTP requests.

On the backend

• The backend has been updated to make use of Microsoft's Face Recognition
• Using the MS API, 2 methods were implemented on the backend:
• callMsDetect which works to build an msDetectOptions object which the MS API requires to analyze and get data from face images.
• The function signature is as follows:

function callMsDetect(imageData, onSuccess, onError)

• The detect options object can be seen below:

    let msDetectOptions = {
        host: config.FACE_API_HOST,
        method: 'POST',
        port: 443,
        path: config.FACE_API_PATH_DETECT,
        headers: {
            'Content-Type': 'application/octet-stream',
            'Content-Length': Buffer.byteLength(imageData),
            'Ocp-Apim-Subscription-Key': config.FACE_API_KEY
        }
    };

• We then make a request to the MS API using the HTTP protocol as shown below:

    let msDetectReq = https.request(msDetectOptions, function(msDetectResponse) {
        msDetectResponse.setEncoding('utf8');
        msDetectResponse.on('data', function(msDetectData){
            onSuccess(JSON.parse(msDetectData));
        });
    });

• On success, we write our Image Data and return that as the result and on error, we return the aforementioned error.
• This method takse the imageData from the frontend and two other callback functions for success and errors if any.

*callMsCompare which works to compare 2 faces based on their data and determine if they are the same or similar to each other.

• This method's signature is as shown below:

function callMsCompare(faceId1, faceId2, onSuccess, onError))

• The verify options object is as follows:

    let msVerifyOptions = {
        host: config.FACE_API_HOST,
        method: 'POST',
        port: 443,
        path: config.FACE_API_PATH_VERIFY,
        headers: {
            'Ocp-Apim-Subscription-Key': config.FACE_API_KEY
        }
    }

• We make a request to the MS API as follows:

    let msVerifyReq = https.request(msVerifyOptions, function(msVerifyResponse) {
        msVerifyResponse.setEncoding('utf8');
        msVerifyResponse.on('data', function(msVerifyData) {
            onSuccess(JSON.parse(msVerifyData));
        });
    })

• More on the Microsoft Face Recognition software can be found here