Implement SAST resizer from FUNQUE #1361
Open
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
The FUNQUE implementation at https://github.com/cosmin/vmaf/tree/funque contains an optional SAST resizer which downscales the input image dimensions by a factor of 2.
This brings in that implementation of the resizer with an optimized code path for 0.5x downscaling with AVX2, AVX512 and NEON SIMD implementation.
In theory this could be used by more than just the FUNQUE extractor, for example SSIM can benefit from the same SAST transform[1] and we could use this for an optimized standalone eSSIM feature extractor.
[1] A. K. Venkataramanan, C. Wu, A. C. Bovik, I. Katsavounidis and Z. Shahid, "A Hitchhiker’s Guide to Structural Similarity," in IEEE Access, vol. 9, pp. 28872-28896, 2021, doi: 10.1109/ACCESS.2021.3056504 https://ieeexplore.ieee.org/document/9344648