TheR1sing3un opened a new pull request, #10096: URL: https://github.com/apache/paimon/pull/10096
### Purpose Decode palette image colors before converting image bytes to training tensors. A red/green palette PNG previously produced a single channel containing palette indices `[0, 1]`, rather than RGB pixels; the shared LeRobot reader could also reject the resulting channel count. Expand palette images to RGB, or RGBA when transparency is present, after applying EXIF orientation. Grayscale and higher-bit-depth conversion retain their existing behavior. ### Tests - All 8 image/numeric window-transform tests passed with Pillow 12.3.0 and Torch 2.8.0. - New regressions verify exact RGB colors, transparent palette index zero, per-entry alpha values, float/uint8 output, the shared LeRobot image path, and mixed palette/RGB windows with EXIF orientation. - New regressions fail against the original implementation. Existing grayscale and 16-bit pixel tests pass. - Python 3.6 syntax, flake8, license headers and `git diff --check` passed. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
