liteflownet-tf2
liteflownet-tf2 copied to clipboard
`tf_warp` generates different results than `tfa.dense_image_warp()`
Based on one simple test I found that the tf_warp(image, flow)
produced different result than tfa.dense_image_warp(image, flow)
. (https://www.tensorflow.org/addons/api_docs/python/tfa/image/dense_image_warp)
Would you please verify if they should be producing the same results? Also would you explain why did you choose to write your own warping function instead of using the tfa one? Thanks!