When running the following commands:
./LaBGen-OF-cli -i input/ -o output/dis_flow.png -a dis_flow -p 1 -n 3 -s 10 -t 2 -r record/dis_flow.avi 1
./LaBGen-OF-cli -i input/ -o output/sparse_to_dense.png -a sparse_to_dense -p 1 -n 6 -s 10 -t 3 -r record/sparse_to_dense.avi 1
The resulting output image is identical if the set of input images were identical too. I have tested this on a large dataset, with all the optical flow implementations provided in this repository. Behavior is only observed between dis_flow and sparse_to_dense.
Note, in the recording it is observed that the vector field, optical flow (hsv) and segmentation map are vastly different. It are merely the output images that are identical.
When running the following commands:
./LaBGen-OF-cli -i input/ -o output/dis_flow.png -a dis_flow -p 1 -n 3 -s 10 -t 2 -r record/dis_flow.avi 1
./LaBGen-OF-cli -i input/ -o output/sparse_to_dense.png -a sparse_to_dense -p 1 -n 6 -s 10 -t 3 -r record/sparse_to_dense.avi 1
The resulting output image is identical if the set of input images were identical too. I have tested this on a large dataset, with all the optical flow implementations provided in this repository. Behavior is only observed between
dis_flowandsparse_to_dense.Note, in the recording it is observed that the vector field, optical flow (hsv) and segmentation map are vastly different. It are merely the output images that are identical.