I tried to capture (stereo) images with Python's opencv and two cameras so therefore every 5 seconds an image should be saved. But the problem here is that an old frame is saved.
The minified code is as follows:
cap = cv2.VideoCapture(0)
for i in range(20):
time.sleep(5)
print "Taking image %d:" % i
ret, frame = cap.read()
cv2.imwrite("image %d" % i, frame)
print " image done." if ret else " Error while taking image..."
cap.release()
cv2.destroyAllWindows()
To check this, I changed the position of the camera after each taken image. But nevertheless an image from an old position is saved (actually not the same, but I assume some frames after the last saved image). After 5 (or more) images finally the captured location in the image does also change.
So, is there any problem with time.sleep? I guess that I'm not getting the actual frame, but a buffered one. If this is the case, how could I fix it and capture the actual frame?
VideoCapture is buffering.
If you always want the actual frame, do this:
while True:
cap = cv2.VideoCapture(0)
ret, frame = cap.read()
cap.release()
cv2.imshow(" ", frame)
if cv2.waitKey(2000) != -1:
break
you need to count the elapsed time, but not stop read frames. like this:
import cv2
import time
cap = cv2.VideoCapture(0)
preframe_tm = time.time()
i = 0
while True:
ret, frame = cap.read()
elapsed_time = time.time() - preframe_tm
if elapsed_time < 5:
continue
preframe_tm = time.time()
i += 1
print("Taking image %d:" % i)
cv2.imwrite("image_%d.jpg" % i, frame)
if i >= 20:
break
cap.release()
cv2.destroyAllWindows()
Related
I have videos I need to grab frames of and store as png images.
I need only frames from specific times.
These times are in microseconds.
How can I grab only these frames?
ret, frame = cap.read()
cv2.imshow("Video", frame)
cap = cv2.VideoCapture("video.mp4")
count = 0
while vidcap.isOpened():
if count == int(308608300 / 1000000):
cv2.imwrite(os.path.join(path_output_dir, '%d.png') % count)
cv2.destroyAllWindows()
This sounds kind of similar with what I'm working on. So for this code just change the number at 'frameRate'.
import cv2
vidcap = cv2.VideoCapture('video.mp4')
def getFrame(sec):
vidcap.set(cv2.CAP_PROP_POS_MSEC,sec*1000)
hasFrames,image = vidcap.read()
if hasFrames:
cv2.imwrite("folder/"+str(sec)+" sec.png", image) # save frame as PNG file
return hasFrames
sec = 0
frameRate = 0.25#it will capture image in each 0.25 second
success = getFrame(sec)
while success:
sec = sec + frameRate
sec = round(sec, 2)
success = getFrame(sec)
I have the code to get images from the video stream of a laptop camera. I want to reduce the photo saving interval to one photo per minute. The original code looks like this
# Importing all necessary libraries
import cv2
import os
# Read the video from specified path
cam = cv2.VideoCapture(0)
try:
# creating a folder named data
if not os.path.exists('data'):
os.makedirs('data')
# if not created then raise error
except OSError:
print ('Error: Creating directory of data')
# frame
currentframe = 0
while(True):
# reading from frame
ret,frame = cam.read()
if ret:
# if video is still left continue creating images
name = './data/frame' + str(currentframe) + '.jpg'
print ('Creating...' + name)
# writing the extracted images
cv2.imwrite(name, frame)
# increasing counter so that it will
# show how many frames are created
currentframe += 1
else:
break
# Release all space and windows once done
cam.release()
cv2.destroyAllWindows()
For this task I try to use the parameter CAP_PROP_POS_MSEC
[...]
# Read the video from specified path
cam = cv2.VideoCapture(0)
cam.set(cv2.CAP_PROP_POS_MSEC,20000)
[...]
while(True):
[...]
# writing the extracted images
cv2.imwrite(name, frame)
cv2.waitKey()
[...]
But, the saving speed remains the same and I see the following error
videoio error v4l2 property pos_msec is not supported
I use Ubuntu 18.04, Python 3.7, and OpenCV 4.1.
Where do I have a mistake, and whether I chose the right way to minimize the load on my computer's resources?
UPD
Using the recommendation of J.D. this code is working
import cv2
import os
import time
prev_time = time.time()
delay = 1 # in seconds
# Read the video from specified path
cam = cv2.VideoCapture(0)
currentframe = 0
while (True):
# reading from frame
ret, frame = cam.read()
if ret:
if time.time() - prev_time > delay:
# if video is still left continue creating images
name = './data/frame' + str(currentframe) + '.jpg'
print('Creating...' + name)
# writing the extracted images
cv2.imwrite(name, frame)
currentframe += 1
prev_time = time.time()
else:
break
EDIT: this answer is not a good solution - due to the frame buffer, as described in the comments. Because of the relevant information in the comments I will leave the answer.
If you don't plan to expand the code to do other things, you can just use the waitkey:
cv2.waitKey(60000) will freeze code execution 60 secs.
If you want to expand the code, you have to create a time based loop:
import time
prev_time = time.time()
count = 0
delay = 1 # in seconds
while True:
if time.time()-prev_time > delay:
count += 1
print(count)
prev_time = time.time()
you should read separately and set the values again for each image, like this:
while(True):
cam = cv2.VideoCapture(0)
cam.set(cv2.CAP_PROP_POS_MSEC,20000)
# reading from frame
ret,frame = cam.read()
if ret:
# if video is still left continue creating images
name = './data/frame' + str(currentframe) + '.jpg'
print ('Creating...' + name)
# writing the extracted images
cv2.imwrite(name, frame)
cv2.waitKey()
# increasing counter so that it will
# show how many frames are created
currentframe += 1
else:
break
cam.release()
and check your openCV version for python about the error.
I know there are many similar questions asked in 2012 or something.
It is 2019, and in 2012, the only solution was to store the frames in memory or write them and then read them in reverse order.
But I was wondering, if there is a new function to read OpenCV frames in reverse order.
I am looking for a solution something like this:
if rewind == False:
ret,frame = cap.read()
else:
ret,frame = cap.read(rewind=True)
Any solution similar to this one?
or something like this:
if rewind == False:
ret,frame = cap.read()
else:
cap.goBackOneFrame()
ret,frame = cap.read()
Edit:
I tried
if rewind == False:
ret, frame = cap.read() #Read frame
counter += 1
else:
cap.set(cv2.CAP_PROP_POS_FRAMES, counter-1)
counter -= 1
ret, frame = cap.read()
and it works, but the rewind speed is slow. I think my PC is taking more time in processing it?
import cv2
# Grab the current frame.
my_check , vid = cap.read()
# use counter variable for
# Counting frames
counter = 0
check = True
frame_list = []
while(check == True):
cv2.imwrite("frame%d.jpg" %counter , vid)
check , vid = cap.read()
frame_list.append(vid)
# increment the counter by 1
counter += 1
frame_list.pop()
# looping in the List of frames.
for frame in frame_list:
# show the frame.
cv2.imshow("Frame" , frame)
if cv2.waitKey(25) and 0xFF == ord("q"):
break
cap.release()
# close any open windows
cv2.destroyAllWindows()
frame_list.reverse()
for frame in frame_list:
cv2.imshow("Frame" , frame)
if cv2.waitKey(25) and 0xFF == ord("q"):
break
cap.release()
cv2.destroyAllWindows()
Have attached the source code for frame differentiating and storing the differentiated frames in a specified place but am getting an error in indentation of error..post this problem as a question in stack overflow..i am restricted to question for a particular period..upload the code too
filename.py
import cv2
import os
import glob
def extractFrames(pathIn, pathOut):
os.mkdir(pathOut)
cap = cv2.VideoCapture(pathIn)
count = 0
while (cap.isOpened()):
# Capture frame-by-frame
ret, frame = cap.read()
current_frame_gray = cv2.cvtColor(current_frame, cv2.COLOR_BGR2GRAY)
previous_frame_gray = cv2.cvtcolor(previous_frame, cv2.COLOR_BGR2GRAY)
frame_diff = cv2.absdiff(current_frame_gray,previous_frame_gray)
if ret == True:
print('Read %d frame: ' % count, ret)
cv2.imwrite(os.path.join(pathOut, "frame{:d}.jpg".format(count)), frame_diff) # save frame as JPEG file
count += 1
else:
break
# When everything done, release the capture
cap.release()
cv2.destroyAllWindows()
def main():
extractFrames('C:/Users/yaazmoha/Desktop/BE PROJECT/INPUT/Tiger in field(1080P_HD).mp4', 'fd3')
if __name__=="__main__":
main()
Fixed your code. You had some indentation errors. Since Python does not use braces like C++, it requires proper indentation to separate code.
import cv2
import os
import glob
def extractFrames(pathIn, pathOut):
os.mkdir(pathOut)
cap = cv2.VideoCapture(pathIn)
count = 0
while (cap.isOpened()):
# Capture frame-by-frame
ret, current_frame = cap.read()
current_frame_gray = cv2.cvtColor(current_frame, cv2.COLOR_BGR2GRAY)
if count > 1:
previous_frame_gray = cv2.cvtcolor(previous_frame, cv2.COLOR_BGR2GRAY)
frame_diff = cv2.absdiff(current_frame_gray,previous_frame_gray)
if ret == True:
print('Read %d frame: ' % count, ret)
cv2.imwrite(os.path.join(pathOut, "frame{:d}.jpg".format(count)), frame_diff) # save frame as JPEG file
count += 1
else:
break
previous_frame = current_frame
# When everything done, release the capture
cap.release()
cv2.destroyAllWindows()
def main():
extractFrames(r"C:\Users\mathesn\Downloads\Wildlife.mp4", 'fd3')
if __name__=="__main__":
main()
I took liberties to fix other sections of your code. But there are some other fixes that this code needs, like creating a directory only if it doesn't exist, maintaining a colored version of the frame so that cv2.cvtColor() does not fail, etc., but I''ll leave those to you.
I am trying to pull individual frames at specified times from an RTSP feed.
This works fine for video streaming:
vcap = cv2.VideoCapture(RTSP_URL)
while(1):
ret, frame = vcap.read()
cv2.imshow('VIDEO', frame)
cv2.waitKey(1)
But if I want to take an image every second and save it by doing something like this:
vcap = cv2.VideoCapture(RTSP_URL)
for t in range(60):
ret, frame = vcap.read()
if ret:
cv2.imwrite("{}.jpg".format(t), frame)
time.sleep(1);
Every image will look exactly the same as the first image. In every instance ret == True.
(Also this was working fine for me a week ago and then ipython did something that required me to do a re-install)
cv2.waitKey(1000) wouldn't do anything if you didn't show image with cv2.imshow(). Try:
vcap = cv2.VideoCapture(RTSP_URL)
for t in range(60):
ret, frame = vcap.read()
cv2.imwrite('{}.jpg'.format(t), frame)
# this will activate the waitKey funciton
cv2.imshow('preview', frame)
cv2.waitKey(1000)
On another note, iPython/jupyter doesn't play well with the cv2's imshow and the whole GUI functionality. If, for example, you can't break the loop by keypress
if (cv2.waitKey(1000) == 27 & 0xff): break;
Alright, after endless messing with it over the last few days, 1 second is not fast enough for the feed for whatever reason.
This will work:
vcap = cv2.VideoCapture(RTSP_URL)
for t in range(60):
ret, frame = vcap.read()
if ret and t % 1000 == 0:
cv2.imwrite("{}.jpg".format(t), frame)
time.sleep(0.001)