twhui/LiteFlowNet

input data is color or gray image

Closed this issue · 2 comments

./tools/convert_imageset_and_flow.cpp show that you load color image, but as i known, the tranditional optical flow estimation alg is based on gray scale image,why do you use color image?

twhui commented

Color images contain more information than gray-scale images. Most modern optical flow algorithms (e.g. Brox et al.) use color image pair. Of course, you can also use gray-scale images.