Both Ambience 2024.0.16 and Repertoire 2024.0.16 server builds were shipped with additional 49 deep learning jars. Users can either refer to the following list to remove them or leave them as-is in their installation. Removal frees up roughly 466 MB of disk space and the default Ambience 2024.0.16 and Repertoire 2024.0.16 server builds do not depend on them.
This only applies to Ambience/Repertoire 2024.0.16, it does not affect Ambience/Repertoire versions 2024.0.15 and below.
Background
The 2024.0.16 builds carry a larger-than-usual zip size because the deeplearning2 jars; the JavaCPP-presets native/computer-vision stack plus its wrappers were packaged into the main build’s \lib. The jars relate to javaCV, onnxRuntime and opencv-platform, and pull in the full set of JavaCPP preset and per-platform native jars behind them. Removing them reclaims roughly 466 MB disk space and does not affect base build functionality.
1. Confirm the affected build
- Refer to remove-list.zip (518 Bytes) for the 49 jar filenames to delete from the
\libfolder. The list applies to both Ambience-2024.0.16 and Repertoire-2024.0.16. - Open the build’s
\libfolder. - If you see jars named
javacv-*.jar,onnxruntime-*.jar,opencv-*.jar,openblas-*.jarand similar, the build is affected.
2. Review the removal list
The 49 jars fall into these libraries (native per-platform variants included): javacv, javacpp, javacpp-platform, onnxruntime, opencv, opencv-platform, openblas, openblas-platform, ffmpeg, tesseract, leptonica, artoolkitplus, flycapture, videoinput, libdc1394, libfreenect, libfreenect2, librealsense, librealsense2.
Configuration Note: do not remove apache5-client-*.jar. It looks unfamiliar next to the 2024.0.15 (or earlier) apache-client-*.jar, but it is simply the renamed AWS SDK HTTP-client module and not a deep learning library.
3. Delete the listed jars from \lib
Place remove-list.txt beside the build folder and run the removal from within the \lib directory.
In PowerShell, e.g.:
# run from the ...\Repertoire-2024.0.16\lib (or ...\Ambience-2024.0.16\lib) folder
Get-Content "..\remove-list.txt" | ForEach-Object { Remove-Item $_ -ErrorAction SilentlyContinue }
On Linux / macOS from the lib folder, e.g.:
while read -r j; do rm -f "$j"; done < ../remove-list.txt
4. Verify the result
Confirm the count dropped by 49 and the flagged jars are gone, e.g.:
# from the lib folder
ls *.jar | wc -l
ls javacv-*.jar onnxruntime-*.jar opencv-*.jar 2>/dev/null # should list nothing
The jar count in \lib should drop by exactly 49, reclaiming ~466 MB:
- Repertoire-2024.0.16:
\libdrops from 398 to 349 (matching 2024.0.15). - Ambience-2024.0.16:
\libdrops from 478 to 429.
Re-zip the build and the size returns to the expected range.
Note: Upgrading to 2024.0.17 in future removes the need for this step entirely. For enquiries on Ambience 202x with deep learning features, please contact sales@elixirtech.com.