A public Python implementation for generating Dynamic Images introduced in 'Dynamic Image Networks for Action Recognition' by Bilen et al.
Python implementation of the dynamic image technology discussed in 'Dynamic Image Networks for Action Recognition' by Bilen et al. Their paper and GitHub can be found as follows:
Clone the directory, and install the module and it's pre-requisites by running:
python setup.py install
import glob
import cv2
from dynamicimage import get_dynamic_image
def main():
frames = glob.glob('./example_frames/*.jpg')
frames = [cv2.imread(f) for f in frames]
dyn_image = get_dynamic_image(frames, normalized=True)
cv2.imshow('', dyn_image)
cv2.waitKey()
if __name__ == '__main__':
main()
Source Video: https://www.youtube.com/watch?v=fXMDubfvoQE