recognize.py

"""
FINDING PRESIDENT BIDEN
Author: Jim Duran
Date: 2021-03

Joe Biden's Celebrity ID is 3Zw7fr

This script expects to see a directory of frames, one per minute, of a channel feed.
It takes each frame and uses facial recognition to find celebrities.
If the celebrity is Joe Biden, ID 3Zw7fr, it highlights the frame.

"""

import boto3
import os
import datetime as dt
import shutil
from time import sleep, strftime
from os import path

s3 = boto3.resource('s3')

#Input folder will include 1440 frames, one frame per minute
sourceFolder = '../../../../../../Volumes/VTNA_Second_Stream/CNN/frames/'
outputFolder = 'output/'


celebrityID = '3Zw7fr' #Default is Biden's celebrity ID, 3Zw7fr
bidenFrames = [] #empty variable

def find_joe_biden(photo):
    client=boto3.client('rekognition')
    status = ''
    area = 0
    otherCelebs = []
    data = []
    frameDest = ''
    try:
        with open(photo, 'rb') as image:
            response = client.recognize_celebrities(Image={'Bytes': image.read()})
    except FileNotFoundError:
        issueFrame = photo.replace('../../../../../../Volumes/VTNA_Second_Stream/CNN/frames/','')
        print('Trouble finding a frame')
        sleep(30)
        data = ['error','error']
        return data

    for celebrity in response['CelebrityFaces']:
        if celebrity['Id'] ==  '3Zw7fr': #Biden's celebrity ID
            area = (float(celebrity['Face']['BoundingBox']['Width'])+float(celebrity['Face']['BoundingBox']['Height']))/2
            """
            The area is the size of the celebrity's face on the screen as a ratio of total screen size.
            0.06 is the size of a little thumbnail in the corner
            0.15 or greater is a head and shoulder shot
            """
            if area >= 0.12:
                status = ">>>>>>------>>>>>>       PRESIDENT BIDEN      <<<<<<-------<<<<<<"
                frameDest = photo.replace(sourceFolder,'')
                shutil.copy2(photo,outputFolder+frameDest)
        else:
            otherCelebs.append(celebrity['Name'])
    data.append(status)
    fullCelebList = ";".join(otherCelebs)
    data.append(fullCelebList)
    return data

def reportWriter(data,date):
    f = open('output/Biden_Report_'+str(date)+'.txt','w')
    f.write('Title:BIDEN REPORT '+str(date)+'\n')
    for item in data:
        f.write(str(item[0]+item[1]+'    '+item[2])+'\n')
    f.close()
    print("File Saved")
    return

def timeIntrepreter(filename):
    hour = int(filename[13:15])
    minutes = int(filename[20:23])
    addhours, min = divmod(minutes,60)
    finalHour = hour + addhours
    return "%d:%02d    " % (finalHour,min)


def main(date):
    ctr = 0
    blocks = ['_0000img','_0400img','_0800img','_1200img','_1600img','_2000img']
    bidenFrames = []
    for block in blocks:
        blockPrefix = 'CNN_' + date + block
        frameList = []
        i = 1
        while i <= 250: #2 for testing, 250 for production
            tempFileName = blockPrefix+str(i).zfill(3)+'.jpg'
            print(tempFileName)
            frameList.append(tempFileName)
            i += 1
        for photo in frameList:
            if '.jpg' in photo:
                print(photo)
                timestamp = timeIntrepreter(photo)
                if path.exists(sourceFolder + photo):
                    photoReport = find_joe_biden(sourceFolder + photo)
                    bidenFrames.append([timestamp,photoReport[0],photoReport[1]])
                    if photoReport[0] == ">>>>>>------>>>>>>       PRESIDENT BIDEN      <<<<<<-------<<<<<<":
                        ctr += 1
                else:
                    photoReport = "missing frame"
                    bidenFrames.append([timestamp,photoReport,'none'])
                    print('does not exist')
                    continue
        print("finished "+block+" Length of frames:"+str(len(bidenFrames)))
    print(bidenFrames)
    print("Biden Count:"+str(ctr))
    return bidenFrames

if __name__ == "__main__":
    print("running main program")
    date = input('select a date YYYYMMDD:')
    report = main(date)
    reportWriter(report,date)