Face-detection-using-MTCNN

face detection using mtcnn complete implementation step by step

Installation

Python version should be 3.7. Run the following commands one by one sequently.

 pip install opencv-python==4.4.0.46
 
 pip install keras==2.3.1
 
 pip install tensorflow==2.2.1
 
 pip install mtcnn==0.1.0
 
 pip install numpy==1.16.6

Face Detection in Image

Run the "mtcnn_image.py" file and keep your input image in the "input & output" folder and set your image path (or name) in the "mtcnn_image.py" file. The output save in the "input & output" folder with "Outputfile.jpg" name (you can also change output file name). Test image is shown below:

Face Detection in .MP4 file

Run the "mtcnn_video.py" file and keep your input video in the "input & output" folder and set your image path (or name) in the "mtcnn_video.py" file. The output save in the "input & output" folder with "filename.avi" name (you can also change output file name). Test Video is shown below:

Face Detection using live camera

Run the "mtcnn_live_cam.py" file or run the following code for live face detection:

import cv2
from mtcnn.mtcnn import MTCNN
detector = MTCNN()

video = cv2.VideoCapture(0)

if (video.isOpened() == False):
    print("Web Camera not detected")
while (True):
    ret, frame = video.read()
    if ret == True:
        location = detector.detect_faces(frame)
        if len(location) > 0:
            for face in location:
                x, y, width, height = face['box']
                x2, y2 = x + width, y + height
                cv2.rectangle(frame, (x, y), (x2, y2), (0, 0, 255), 4)
        cv2.imshow("Output",frame)
        if cv2.waitKey(1) & 0xFF == ord('q'):
            break
    else:
        break

video.release()
cv2.destroyAllWindows()

Live Camera Output