detector.py

__author__ = 'Brian Diaz'
import cv2
from face import Face
from eye import Eye
from classifier import Classifier
import numpy as np


class Detector:

	def __init__(self, image, scale_factor, minimum_neighbors, minimum_size, flags, classifier):
		self.face_cascade = classifier.face_cascade
		self.eye_cascade = classifier.eye_cascade
		self.smile_cascade = classifier.smile_cascade
		self.image = image
		self.grayscale_image = self.__convert_image_to_gray(image)
		self.scale_factor = scale_factor
		self.minimum_neighbors = minimum_neighbors
		self.minimum_size = minimum_size
		self.flags = flags

	def detect(self):
		faces = []
		eyes = []
		rgb_green = (0, 255, 0)
		rgb_blue = (0, 0, 255)

		detected_faces = self.__detect_faces()

		for (x, y, w, h) in detected_faces:
			face = Face(self.image, x, y, w, h, rgb_green)
			face.draw(2)
			faces.append(face)
			detected_eyes = self.__detect_eyes(face.data())
			if len(detected_eyes) > 0:
				for(eye_x, eye_y, eye_w, eye_h) in detected_eyes:
					eye = Eye(self.image, face.x + eye_x, face.y + eye_y, eye_w, eye_h, rgb_blue)
					eye.draw(2)
					eyes.append(eye)

			detected_smile = self.__detect_smile(face.data())
			if len(detected_smile) > 0:
				smileness = detected_smile[0][1]
				smile_text_position = ((x - 15), y + (h + 25))
				if smileness >= 15:
					cv2.putText(self.image, "Smiling: Yes", smile_text_position, cv2.FONT_HERSHEY_DUPLEX, 1, (255,255,255))
				else:
					cv2.putText(self.image, "Smiling: No", smile_text_position, cv2.FONT_HERSHEY_DUPLEX, 1, (255,255,255))

			return { "faces" : faces, "eyes" : eyes}

	# Private Methods
	def __convert_image_to_gray(self, image):
		return cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)

	def __detect_faces(self):
		faces = self.face_cascade.detectMultiScale(
		    self.grayscale_image,
		    scaleFactor=self.scale_factor,
		    minNeighbors=self.minimum_neighbors,
		    minSize=self.minimum_size,
		    flags = self.flags
		)
		return faces

	def __detect_eyes(self, image):
		eyes = self.eye_cascade.detectMultiScale(
		    self.__convert_image_to_gray(image)
		)
		return eyes

	def __detect_smile(self, image):
		smiles = self.face_cascade.detectMultiScale(
		    self.__convert_image_to_gray(image), 
		    scaleFactor=self.scale_factor,
		    minNeighbors=self.minimum_neighbors,
		    minSize=self.minimum_size,
		    flags = self.flags
		)
		return smiles