Add Q-Align splits for KADID10k and SPAQ
#1
by chaofengc - opened
Add a qalign_split column to KADID10k and SPAQ metadata using exact Q-Align test-manifest image membership. Existing HF split columns and all prior metadata values are preserved. The README clarifies that train rows are the complement of the Q-Align test manifest, not confirmed historical Q-Align training usage.
Superseded by PR #2, which moves qalign_split next to the image filename so it is immediately visible in the CSV preview: https://huggingface.co/datasets/chaofengc/IQA-PyTorch-Datasets-metainfo/discussions/2
chaofengc changed pull request status to closed