fullbodydetection.py

##########################################
#****** ML Height Estimate Project ******#
#** Rahul Rachh    - 1401057 **#
#** Kedar Acharya  - 1401081 **#
#** Janki Desai    - 1401103 **#
#** Shloak Agarwal - 1401105 **#
##########################################

#We can initialize numpy arrays from nested Python lists, and access elements using square brackets.
import numpy as np
#Wraps the python api, and uses numpy arrays for images
import cv2
import math

phi = float(input("Please enter value of tilt angle (in radians):")); #tilt angle is angle made by camera to horizontal surface 

#'def' is like a 'function' used in c++.A function is a block of organized, reusable code that is used to perform a single, related action.
#The function name is 'inside' with the parameters 'r' & 'q' and the function returns 'true' and 'false' according to the result.
def inside(r, q):
    #Using the logical operator to equate the vlaue of 'r' and 'q' respectively to variables.
    rx, ry, rw, rh = r
    qx, qy, qw, qh = q
    #def inside returns 'true' or 'false' according to the result of relational operator.
    return rx > qx and ry > qy and rx + rw < qx + qw and ry + rh < qy + qh

#'def' is like a 'function' used in c++.A function is a block of organized, reusable code that is used to perform a single, related action.
#The function name is 'draw_detections' with parameters 'img'(gray scale image/video), 'rects'(rectangle paramters) & thickness(thickness of the border(predefine=2)).
#'draw_detections' draws the rectangle around the human body that is detected with the border thickness = 2.
#Futher the function also detect the height of the human body that is detected.
def draw_detections(img, rects, thickness = 2):
    #'for' loop until the rectangle is displayed.
    for x, y, w, h in rects:
        #The HOG detector returns slightly larger rectangles than the real objects.
        #So we slightly shrink the rectangles to get a nicer output.
        pad_w, pad_h = int(0.1512*w), int(0.0482*h)  #Continuosly experimenting with the values these suggests to be the most accurate one for the measurement of height
        #Forming and displaying the rectangle.
        cv2.rectangle(img, (x+pad_w, y+pad_h), (x+w-pad_w, y+h-pad_h), (0, 255, 0), thickness)
        #'P_ref' stands for pixel reference which is equal to total number of pixels present in height of frame (display resolution of frame is 640X480)
        P_ref = 400
        #'H_ref' stands for height reference which is equal to height of referred person  
        H_ref = 165
        #'P_left' is the pixel that are left between the bottom of the rectangle frame and the bottom of the display screen.
        P_left = float(P_ref-float((y+h-pad_h)/math.cos(phi)));
        h = float(h/math.cos(phi));
        if(P_left<200):
            Height = float(((h-(2*pad_h))*H_ref)/(P_ref - (2*P_left)));#Predicted height according to the formula.
            
            P_height = int(Height);
            if(P_height>100 ):
                text_color = (0,255,0)
                if(phi != 1.57 ):
                    cv2.putText(resized_image, str(P_height)+"cm" , (x+pad_w, y+pad_h), cv2.FONT_HERSHEY_PLAIN, 1.2, text_color, thickness=2)
                    print float(P_height);
            print "Height:" + str(P_height);

#While running the python file name is changed to main

if __name__ == '__main__':

    hog = cv2.HOGDescriptor()
    hog.setSVMDetector( cv2.HOGDescriptor_getDefaultPeopleDetector() )
    cap=cv2.VideoCapture(0)# capture real time video from camera
    #for seg
    fgbg = cv2.createBackgroundSubtractorMOG2()
    
    while True:
        ret,frame=cap.read()
        #for grayscale
        gray = cv2.cvtColor(frame, cv2.COLOR_BGR2GRAY)
        #fgmask = fgbg.apply(frame)       
        resized_image = cv2.resize(gray, (400, 400))#image resizing to reduce computational complexity and to increase FPS
        found,w=hog.detectMultiScale(resized_image, winStride=(8,8), padding=(32,32), scale=1.05)
        draw_detections(resized_image,found)
        cv2.imshow('feed',resized_image)
        
        ch = 0xFF & cv2.waitKey(1)
        if ch == 27:
            break
    
    cv2.release()
    cv2.destroyAllWindows()