app.py

from flask import Flask, render_template, Response, jsonify, request
import cv2
import os
import requests
from PIL import Image
from script import predict
import numpy as np
from evaluate import execute
from pose_parser import pose_parse
import pickle
from sklearn.cluster import KMeans
import description
import time
cascPath = "./models/haarcascade_frontalface_default.xml"
faceCascade = cv2.CascadeClassifier(cascPath)
app = Flask(__name__)
flag = True
name = ""
selection = ""
title = ""
price = ""
kmeansPath = "./models/kmeans.pkl"
with open(kmeansPath, 'rb') as file:
    kmeans = pickle.load(file)


class VideoCamera(object):
    def __init__(self):
        self.video = cv2.VideoCapture(0)

    def __del__(self):
        self.video.release()

    def get_frame(self):
        ret, frame = self.video.read()
        frame = frame[48:433, 176:465]
        img = cv2.resize(frame, (192, 256))
        cv2.imwrite('temp.jpg', img)
        gray = cv2.cvtColor(frame, cv2.COLOR_BGR2GRAY)
        faces = faceCascade.detectMultiScale(
            gray,
            scaleFactor=1.1,
            minNeighbors=5,
            minSize=(30, 30)
        )
        err = False
        if len(faces) == 0:
            err = True
        for (x, y, w, h) in faces:
            cv2.imwrite('face.jpg', frame[y-10:y+h+10, x:x+w])
            cv2.rectangle(frame, (x, y), (x+w, y+h), (0, 255, 0), 2)

        return frame, err


@app.route('/')
def index():
    global flag
    flag = True
    return render_template('index.html')


@app.route('/final')
def final():
    return render_template('final.html')


@app.route('/men')
def men():
    return render_template('men.html')


@app.route('/women')
def women():
    return render_template('women.html')


@app.route('/feature', methods=['GET', 'POST'])
def feature():
    if request.method == "GET":
        img1 = open('face.jpg', 'rb').read()
        headers = {'Ocp-Apim-Subscription-Key': "//key",
                   'Content-Type': 'application/octet-stream'}
        response = requests.post(
            "https://cv21.cognitiveservices.azure.com/vision/v3.1/analyze?visualFeatures=faces,color", headers=headers, data=img1)
        response.raise_for_status()
        analysis = response.json()
        try:
            age = analysis["faces"][0]["age"]
        except:
            age = 23
        age = int(age)-10
        try:
            gender = analysis["faces"][0]["gender"]
        except:
            gender = "Male"
        try:
            ethnicity = analysis["color"]["dominantColors"][0]
        except:
            ethnicity = "Brown"
    else:
        age = int(request.form["age"])
        gender = request.form["gender"]
        ethnicity = request.form["eth"]
    print(ethnicity)
    a = 100 if gender == "Female" else -100
    b = age-2
    c = age+2
    if ethnicity == "White" or ethnicity == "Pink":
        d, e, f = 1, 0, 0
    elif ethnicity == "Black":
        d, e, f = 0, 1, 0
    else:
        d, e, f = 0, 0, 1
    cluster = kmeans.predict(np.array([[a, b, c, d, e, f]]))[0]
    print(cluster)
    clusters = os.listdir('./clusters/'+str(cluster))
    return render_template('projection.html', age=age, gender=gender, eth=ethnicity, cluster=clusters, len=len(clusters), desc=description.desc[cluster])


@app.route('/cast', methods=['POST'])
def cast():
    global name, selection, title, price
    name = request.form["name"]
    selection = request.form["selection"]
    title = request.form["title"]
    price = request.form["price"]
    title = title.strip()
    return render_template('final.html', sel=selection, title=title, price=price)


def gen(camera):
    timer = 100
    global video_stream
    while timer > 0:
        frame, err = camera.get_frame()
        frame = cv2.putText(frame, str(timer//10), (20, 40), cv2.FONT_HERSHEY_SIMPLEX,
                            1.5, (0, 0, 0), 4, cv2.LINE_AA)
        ret, jpeg = cv2.imencode('.jpg', frame)
        frame = jpeg.tobytes()
        if err != True:
            timer -= 1
        yield (b'--frame\r\n'
               b'Content-Type: image/jpeg\r\n\r\n' + frame + b'\r\n\r\n')
    video_stream.__del__()


def gen_stored(path):
    img = cv2.imread(path)
    ret, jpeg = cv2.imencode('.jpg', img)
    frame = jpeg.tobytes()
    yield (b'--frame\r\n'b'Content-Type: image/jpeg\r\n\r\n' + frame + b'\r\n\r\n')


@app.route('/video_feed')
def video_feed():
    global video_stream, flag
    if flag:
        video_stream = VideoCamera()
        flag = False
        return Response(gen(video_stream), mimetype='multipart/x-mixed-replace; boundary=frame')
    else:
        return Response(gen_stored("temp.jpg"), mimetype='multipart/x-mixed-replace; boundary=frame')


@app.route('/final_img')
def final_img():
    global name, selection
    person = Image.open('temp.jpg')
    person.save("./Database/val/person/"+name+".jpg")
    pose_parse(name)
    execute()
    f = open("./Database/val_pairs.txt", "w")
    f.write(name+".jpg "+selection+"_1.jpg")
    f.close()
    predict()
    im = Image.open("./output/second/TOM/val/" + selection + "_1.jpg")
    width, height = im.size
    left = width / 3
    top = 2 * height / 3
    right = width
    bottom = height
    im1 = im.crop((left, top, right, bottom))
    newsize = (600, 450)
    im1 = im1.resize(newsize)
    im1.save("./output/second/TOM/val/" + selection + "_1.jpg")
    result = Image.open("./output/second/TOM/val/" + selection + "_1.jpg")
    result.save("data.jpg")
    return Response(gen_stored("data.jpg"), mimetype='multipart/x-mixed-replace; boundary=frame')


if __name__ == '__main__':
    app.run(host='localhost', port=5000, debug=False)