Just read two images(like two frames), then go through the image diff, gaussian blur, canny edge detection, dilate, and erode, to get some little spots. Then another program perhaps can use their result to find the moving targets.
liuyh-horizon / canny2hell_vitis_aws_xacc2021 Goto Github PK
View Code? Open in Web Editor NEWLicense: GNU General Public License v3.0