I am new computer vision and I am having a dataset with consecutive frames of videos. I want to ask that is it okay? How will you deal with it to output a video with some detections?
1
-
$\begingroup$ Please mention more details. What is the task you are trying to solve? What are you concerned about? $\endgroup$– Devashish PrasadJun 3, 2021 at 19:58
Add a comment
|
2 Answers
$\begingroup$
$\endgroup$
Yes - individual frames can be used. See existing answers and links to docs as appropriate:
https://stackoverflow.com/q/18954889/1928322
The following blog shows how to extract individual frames from a video (replicating your environment) before continuing: https://www.pyimagesearch.com/2017/02/06/faster-video-file-fps-with-cv2-videocapture-and-opencv/
$\begingroup$
$\endgroup$
Video has two information i.e. Images(Frames) and a temporal sense(moving Frame).
- To identify an object in each frame, treating each frame as separate image will work.
- To read the temporal sense of the video i.e. Running or Skipping, will need to treat the video as Sequence of images and work accordingly.
This topic is pretty well covered in the online course of "University of Michigan"
Course Link
Check lecture 17,18.
