Is there a way to export/checkpoint OpenCV Background Subtraction for later use?

Question

I have some very long video files to process which require background removal. I would like to cut the video into small chunks and process each chunk separately. However, in doing this I would need to generate a new background subtraction model for each small video chunk. This would eat into my usable data in these very long videos.

Can I checkpoint this black box somehow? If so, how can I start up a video with this exported checkpoint information?

Version Info

Python 3.6.4
OpenCV 3.4.1

(Answers from C++ users welcome, though Python is preferred)

MWE

import numpy as np

import cv2

FGBG = cv2.bgsegm.createBackgroundSubtractorMOG(100, 7, 0.5, 5)
MAT = np.ones((3, 3), np.uint8)
pos_frame = 0

while pos_frame < 1000:
    cap = cv2.VideoCapture(the_file)
    cap.set(cv2.CAP_PROP_POS_FRAMES, pos_frame)
    ret, frame = cap.read()
    if ret:
        frame = cv2.cvtColor(frame, cv2.COLOR_BGR2GRAY)
        frame = cv2.GaussianBlur(frame, (5, 5), 0)
        frame = FGBG.apply(frame, learningRate=.05)
        frame = cv2.morphologyEx(frame, cv2.MORPH_GRADIENT, MAT)
        cv2.waitKey(1)
    pos_frame += 1

# This is a placeholder for what I want to do.
cv2.somehowexportthebackgroundinfofromabove()

Is there a way to export/checkpoint OpenCV Background Subtraction for later use?

0 Answers0