This script requires IPCam app (or any other app that can stream images/videos over HTTP) to be installed on Android device and OpenCV to installed on Desktop machine. On running this script, OpenCV will detect front-faces in videos streamed on the desktop. #python #opencv #android #face-detection

android_video_stream_to_opencv.py

import requests
import cv2
import numpy as np

# Online code reference: https://www.youtube.com/watch?v=-mJXEzSD1Ic

# Image URL being streamed from IPWebCam app (I used the one written by Pavel Khlebovich).
# Make sure that the android device and the machine running opencv are on the same network and
# you can access the streaming URL from that machine
url = 'http://192.168.43.236:8080/shot.jpg'
# HAARCascade Classifier for Frontal faces
face_cascade = cv2.CascadeClassifier("data\haarcascade_frontalface_default.xml")

# Run infinite loop (unless terminated by pressing Esc key on OpenCV video window)
while True:
    # Get image from the IPWeb Cam
    img_resp = requests.get(url)
    img_arr = np.array(bytearray(img_resp.content), dtype=np.uint8)
    img = cv2.imdecode(img_arr, -1)

    # Reduce the size of the image to half to increase performance of recognition algorithm 
    # (it will have to interpret less pixels / data
    resize = cv2.resize(img, (int(img.shape[1]/2),int(img.shape[0]/2)))
    
    # There might be multiple faces in the video/image streamed. Play around with 
    # scaleFactor & minNeighbors to get a satisfactory recognition
    faces = face_cascade.detectMultiScale(resize, scaleFactor=1.03, minNeighbors=5)
    
    # Draw rectangle boxes over the identified faces
    for x,y,w,h in faces:
        resize = cv2.rectangle(resize, (x,y), (x+w, y+h), (0, 255, 0), 3)

    # Show stream with rectangle drawn over faces in opencv window
    cv2.imshow("AndroidCam", resize)

    # continue detection until user hits ESC button
    if cv2.waitKey(1) == 27:
        break