How to save only a part of a long video using OpenCV
Trigger
Gufu / Propagation YouTuber [Takashi Tachibana, which increases with "NHK" and doubles with "break"](https://www. synced to youtube.com/watch?v=NcR2d8_iM9o): rolling_eyes: With just the right video material, it seems that opencv can do something such as area division, face recognition, erasing telop, etc. This video has 18 minutes, but I want to use only the first minute when the candidate appears. So, I will try to extract only a part by time with python and OpenCV only.
Original material
[![Destroy NHK![Political Broadcasting] Party to protect the people from NHK [Takashi Tachibana and 3 others] National Proportional District](https://img.youtube.com/vi/iRi4od_Thus/0.jpg)
](http://www.youtube. com / watch? v = iRi4od_Thus)
Method
--Extract fps (frames per second) from the original video
--Calculate the number of frames with start time * fps or end time * fps
I tried to measure with time.time (), but it was more successful to control it based on the number of frames.
Also, because it is the first minute when the candidate appears
--Start time = 30 seconds --End time = 90 seconds
It was made.
development of
import cv2
if __name__ == '__main__':
cap = cv2.VideoCapture('drop_out_nhk.mp4')
cap_width = int(cap.get(cv2.CAP_PROP_FRAME_WIDTH))
cap_height = int(cap.get(cv2.CAP_PROP_FRAME_HEIGHT))
fps = cap.get(cv2.CAP_PROP_FPS)
fourcc = cv2.VideoWriter_fourcc('m','p','4','v')
writer = cv2.VideoWriter('one_minutes.mp4',fourcc, fps, (cap_width, cap_height))
#Start or end time you want to extract
begin = 30
end = 90
for i in range(end * fps):
ret, frame = cap.read()
if ret:
if begin * fps < i:
writer.write(frame)
writer.release()
cap.release()
result
I think this is just one minute.
in conclusion
I thought I would use ʻasyncio`, but I was able to implement it fairly easily. If you change the image size, the codec will make a mistake for some reason.
Referenced link
-Read video from OpenCV-Python file --OpenCV --Load / write video with VideoCapture / VideoWriter.
Recommended Posts