mlp-project/transform_img.py
2024-02-25 21:03:32 +00:00

6 lines
No EOL
337 B
Python

# If we cannot get revpersnet running,
# we still ought to do some sort of information-preserving perspective transformation
# e.g., randomized transformation
# and let transcrowd to crunch through these transformed image instead.
# After training, we obtain the attention map and put it in our paper.
# I just want to get things done...