camera3.py

import cv2, io, base64
from PIL import Image
import threading
from time import sleep
import numpy as np
from GazeTracking.gaze_tracking import GazeTracking
from fer import FER
from Emotionrec import autism_detector,aut_percentage

trainedfacedata = cv2.CascadeClassifier('haarcascade_frontalface_default.xml')
detection = FER()


class VideoCamera3(object):
    def __init__(self):

        self.to_process = []
        self.output_image_rgb = []
        self.output_image_bgr = []
        thread = threading.Thread(target=self.keep_processing, args=())
        thread.daemon = True
        thread.start()


    def process_one(self):
        if not self.to_process:
            return
        input_str = self.to_process.pop(0)
        imgdata = base64.b64decode(input_str)
        input_img = np.array(Image.open(io.BytesIO(imgdata)))
        """ h
        After getting the image you can do any preprocessing here
        """
        # ______________________________________Performing some pre processing______________________________________________

        bgr_image = cv2.flip(input_img, 1)  # Flip the image

        left = 400
        top = 500
        right = 3800
        bottom = 4000



        cropped = bgr_image
        #cv2.imshow("img", cropped)
        # Converting image to grayscale
        #gray = cv2.cvtColor(cropped, cv2.COLOR_BGR2GRAY)
        #cv2.imwrite('cropped_grayscale.jpg', gray)

        # Detecting emotions of the human with FER module and printing the output
        #image_path = cv2.imread('cropped_grayscale.jpg')
        all_emotions = detection.detect_emotions(cropped)
        all_emotion_list = list(all_emotions)
        print(f'***This is the complete list of all emotions in the human:{all_emotion_list}')

        #image_path = cv2.imread('cropped_grayscale.jpg')
        #detection = FER()
        top_emotion = detection.top_emotion(cropped)
        top_emotion_list = list(top_emotion)
        print(f'***This is the dominant emotion in the human: {top_emotion_list}')

        # Converting FER output into list
        converting_to_list = list(top_emotion)
        mood, per = converting_to_list

        # appending the detected mood and its percent dominance into two new lists.
        emotion_detected = [mood]
        per_detected = [per]
        print(
            f'{autism_detector(emotion_detected)}, with {aut_percentage(per_detected)}% of the odd emotion, which shows'
            f' a lack of interest.')




            # We get a new frame from the webcam



        #cv2.destroyAllWindows()
            # cv2.imshow("me5.png", frame)
        rgb_image = cv2.cvtColor(bgr_image, cv2.COLOR_BGR2RGB)# Changing color from bgr to rgb

        # ____________________________________________________________________________________________________________________

        ret, rgb_jpeg = cv2.imencode('.jpg', rgb_image)
        _, bgr_jpeg = cv2.imencode('.jpg', rgb_image)

        self.output_image_rgb.append(rgb_jpeg.tobytes())
        self.output_image_bgr.append(bgr_jpeg.tobytes())

        #plt.scatter(eye_1x_positions, eye_1y_positions)
        #plt.show()
        #sns.set()





    def keep_processing(self):
        while True:
            self.process_one()
            sleep(0.01)

    def enqueue_input(self, input):
        self.to_process.append(input)

    def get_frame(self):
        while not self.output_image_rgb:
            sleep(0.05)
        return self.output_image_rgb.pop(0), self.output_image_bgr.pop(0)





"""
# def stringToImage(base64_string):
#     imgdata = base64.b64decode(base64_string)
#     return np.array(Image.open(io.BytesIO(imgdata)))
"""