Motion-Extraction

With a very simple trick you can extract the motion of a video. Uses python3 and OpenCV to compare current frame of the video to a previous frame to see where changes have been made in the frames (and any changes indicate motion).

Inspired by this video by Posy

NOTE: Recommend to only test on shorter videos (ie: not greater than 5 minutes) -- current program converts in REAL TIME

• IE: 5 hour video = at least 5 hours to convert

How to run this program (Easy Way):

• 1. Make sure you have python 3 downloaded, installed and on path
• 2. Move your target MP4 video into the same directory as the motion_Extraction.py file
• 5. Run the easy install program with the command python Install_and_Run_Program.py --video 'yourVideo.mp4'
  • you NEED to provide the video title as an argument
  • this will automatically create a new virtual environment, install all the pip packages and then run the extraction program
• 6. If you want to see the extraction overlay, type yes. If you do not, type no
• 7. If you just want the frame compared to the previous one, then type PREVIOUS. If you want a larger delay, type CUSTOM (you will the be prompted to enter your custom frame delay)
• 8. Your output will be created in a subfolder within the OUTPUTS folder, enjoy!

How to run this program (Long Way):

• 1. Make sure you have python 3 downloaded, installed and on path
• 2. Use the activateVirEnv.txt to create a virtual environment
• 3. Make sure you pip install -r requirements.txt to install all the needed packages
• 4. Move your target MP4 video into the same directory as the motion_Extraction.py file
• 5. Run the program with the command python motion_Extraction.py yourVideo.mp4
  • you NEED to provide the video title as an argument
• 6. If you want to see the extraction overlay, type yes. If you do not, type no
• 7. If you just want the frame compared to the previous one, then type PREVIOUS. If you want a larger delay, type CUSTOM (you will the be prompted to enter your custom frame delay)
• 8. Your output will be created in a subfolder within the OUTPUTS folder, enjoy!

Ideas:

• [DONE] maybe combine the two while loops?
• [DONE] invert then immediatly overlay?
• [DONE] check pervious frame in one video then x frames back in another video (shows slower movement)
• cut videos larger than like 3 minutes into chunks of 1 minutes, store in folder [check multiprocessing attempt folder]
• then use threading to individually loop through and invert videos
• then you can paste it together at the end

Resources used:

• Motion Extraction tutorial (attempt...): https://youtu.be/woj4vfMLpao
• Python OpenCV – cv2.flip() method: https://www.geeksforgeeks.org/python-opencv-cv2-flip-method/
• Adding (blending) two images using OpenCV : https://docs.opencv.org/3.4/d5/dc4/tutorial_adding_images.html
• How to split a video in parts using Python?: https://stackoverflow.com/questions/65570944/how-to-split-a-video-in-parts-using-python
• multiprocessing — Process-based parallelism: https://docs.python.org/3/library/multiprocessing.html