firefly.py

# Federico Tondolo
# Summer 2019 at the Columbia University Medical Center 
import numpy as np
import ffmpy
import cv2
import os
import subprocess

video_length=40
prologue=20
safe_space=20

def main():
    # Creating OUTPUT directory
    os.mkdir('OUTPUT')
    # Loops for every file inside current dir
    for filename in os.listdir('.'):
        # If file is a video file
        if (filename.endswith(".wmv")):
            cap = cv2.VideoCapture(filename)
            # Initialisation of frame_counting variable
            frame_count=0
            # Initialisation of lightbulb activations count
            count=0
            # Initialisation of variable defining the next valid frame
            next_valid=0
            # Loop as long as there's a frame to analyze
            while cap.grab():
                # Adding current frame to frame counter
                frame_count+=1
                # If frame is bit within 21 seconds of first trigger
                if (frame_count>=next_valid):
                    # Initialisation of supporting variables
                    flag, frame = cap.retrieve()
                    # Frame manipulation begins here
                    mask = np.zeros(frame.shape, dtype=np.uint8)
                    # Defining of ROI
                    roi_corners = np.array([[(70,33), (100,25), (165, 45), (165,55)]], dtype=np.int32)
                    # FIlling the ROI
                    ignore_mask_color = (255,)*2
                    cv2.fillPoly(mask, roi_corners, ignore_mask_color)
                    # Applying the mask
                    masked_image = cv2.bitwise_and(frame, mask)
                    # Blurring and brightening the resulting frame
                    blur = cv2.blur(masked_image,(6,4))
                    brightest = cv2.threshold(blur, 200, 255, cv2.THRESH_BINARY)[1]
                    if finder(brightest):
                        count+=1
                        start= (frame_count/30)-prologue
                        next_valid=((start+prologue+video_length)*30)
                        subprocess.call(['ffmpeg', '-i', filename, '-ss', str(start), '-t', str(video_length), ('./OUTPUT/output_'+filename[:-4]+'-'+str(count)+'.wmv')])
         # If file is not a video 
        else:
            continue
        
def finder(im):
    # Set up the detector with necessary parameters
    params = cv2.SimpleBlobDetector_Params()
    params.filterByColor = 1
    params.blobColor=255
    params.filterByArea = 1
    params.maxArea=150
    params.minArea=25
    detector = cv2.SimpleBlobDetector_create(params)
    # Detect blobs
    keypoints = detector.detect(im)
    # Blob detected
    if len(keypoints):
        return 1
    
main()