The faceauth from thebrotherfromasouthernmother

thebrotherfromasouthernmother / faceauth Goto Github PK

View Code? Open in Web Editor NEW

An open source web tool that uses computer vision to fight trolls and keep online communities safe. Built upon Node.js, Python, and Postgres.

JavaScript 53.57% HTML 17.99% Python 28.44%

faceauth's People

Watchers

faceauth's Issues

High runtime on large datasets of banned users

Currently this block of code has a pretty high runtime when the number of banned users is high. In order for this application to work in production will likely need to develop an algorithm for sorting banned users into subfolders based on certain facial features so as to decrease the runtime when the number of banned users is high.

def matchFacesInDirectory(unknown_image):
    isUserBanned = False
    directory = os.fsencode("./banned_users");
    integer_i = 0;
    for file in os.listdir(directory):
        if(integer_i == len(os.listdir(directory))):
            break
        filename = os.fsdecode(file)
        isUserBanned = compareTwoImages(filename, unknown_image)
        if (isUserBanned == True):
            print(isUserBanned)
            break
        integer_i += 1
    if (isUserBanned == False):
        print(isUserBanned)
    return isUserBanned

An idea is to implement a tensorflow ML model to categorize faces.

Recommend Projects