emotiscan.py

#!/usr/bin/env python3
import cv2
import sys
from pprint import pprint
import numpy as np
import keras
from keras.models import load_model
from samplebase import SampleBase
from rgbmatrix import graphics
from rgbmatrix import RGBMatrix, RGBMatrixOptions
import time

# Initialize some variables
face_locations = []
face_encodings = []
face_names = []
process_this_frame = True
#model = load_model("./emotions/fer2013_mini_XCEPTION.102-0.66.hdf5")
model = load_model("./emotions/model_v6_23.hdf5")
#emotion_dict= {'Angry': 0, 'Sad': 5, 'Neutral': 4, 'Disgust': 1, 'Surprise': 6, 'Fear': 2, 'Happy': 3}
emotion_dict= {'>:(': 0, ':(': 5, ':|': 4, ':X': 1, ':O': 6, '<:(': 2, ':D': 3}
#cascPath = "./cascades/haarcascade_frontalface_default.xml"
cascPath = "./cascades/haarcascade_frontalface_alt.xml"
faceCascade = cv2.CascadeClassifier(cascPath)
feed_counter = 0
video_capture = cv2.VideoCapture(0)

# Configuration for the matrix
options = RGBMatrixOptions()
options.rows = 32
options.cols = 32
options.disable_hardware_pulsing = 1
options.chain_length = 1
options.parallel = 1

options.hardware_mapping = 'adafruit-hat'  # If you have an Adafruit HAT: 'adafruit-hat'
#options.pwm_bits = 1
#options.pwm_dither_bits = 2

options.gpio_slowdown = 2
matrix = RGBMatrix(options = options)
canvas = matrix
font = graphics.Font()
#font.CharacterWidth(60)
#font.LoadFont("./fonts/custom4.bdf")
font.LoadFont("./fonts/10x20.bdf")
color = graphics.Color(125, 125, 255)

def facecrop(faces, image):
    for f in faces:
        x, y, w, h = [ v for v in f ]
        cv2.rectangle(image, (x,y), (x+w,y+h), (255,255,255))
        sub_face = image[y:y+h, x:x+w]
        sub_face = cv2.resize(sub_face, (48,48))
        #sub_face = cv2.resize(sub_face, (64,64))
        return sub_face

while True:
    # Capture frame-by-frame
    ret, frame = video_capture.read()

    # Only process every Nth frame 
    # 10 = 25-35% CPU on all cores
    # Ideal number = 20
    if (feed_counter == 20):
        gray = cv2.cvtColor(frame, cv2.COLOR_BGR2GRAY)
        faces = faceCascade.detectMultiScale(
            gray,
            scaleFactor=1.1,
            minNeighbors=5,
            minSize=(30, 30),
            flags=cv2.CASCADE_SCALE_IMAGE
        )
        if len(faces) > 0:
            print("Face detected..")
            face_crop = facecrop(faces,gray)
            if face_crop is not None:
                print("Face cropped..")
                face_image = np.reshape(face_crop, [1, face_crop.shape[0], face_crop.shape[1], 1])
                predicted_class = np.argmax(model.predict(face_image))
                label_map = dict((v,k) for k,v in emotion_dict.items())
                predicted_label = label_map[predicted_class]
                pprint(label_map)
                print(predicted_label)
                canvas.Clear()
                graphics.DrawText(canvas, font, 1, 20, color, predicted_label)
                #time.sleep(0.02)
        feed_counter = 0
    feed_counter += 1

# When everything is done, release the capture
video_capture.release()
cv2.destroyAllWindows()