Releases: nextcloud/recognize
Releases 路 nextcloud/recognize
v3.3.5
Fixed
- Add Migration step: Add index for detections#file_id column
- Run custom ffmpeg install in InstallDeps (Support musicnn,movinet on ARM)
- MusicnnClassifier: Increase WASM timeout
- ClusteringFaceClassifier: Filter out non-straight faces
- Set max execution time for classifier cron job
- ViewAdmin: Don't mark "models still need to be downloaded" as success
- Don't rerun InstallDeps when node binary exists already
v3.3.4
Fixed
- increase install nodejs download timeout to 300s
- Implement incremental clustering Marcel Klehr
- FaceClusterAnalyzer: Increase cluster density
- FaceClusterAnalyzer: Filter out too small faces
- ClusteringFaceClassifier: Increase face recognition score
v3.3.3
Fixed
- Fix bundle size
v3.3.2
Fixed
- Don't allow setting cluster title to the same value as the ID of a cluster without a title
- Classifier#getConvertedFilePath: Don't throw if $preview is false
- Classifier: Don't let images larger than 1024x1024 into the classifier
- Classify command: Only add items to queue if model is enabled
- FaceClusterAnalyzer: Ignore clusters that are larger than what could possibly be the same face (Fixes the mega-cluster issue; make sure to delete mega-clusters after the update)
v3.3.1
Fixed
- Add localizations to the build
v3.3.0
New
- Implement recognize:remove-legacy-tags command
- Implement GPU mode
- Implement recognize:classify command
v3.2.3
Fixed
- Remove SchedulerJob from info.xml
- Classifier: Only apply 8MB rule to image classifiers
- Use regex to catch ignored paths instead of enumerating directories
- Use IgnoreService in FileListener
- Add bg job mode status in admin settings
- Add tensorflow WASM status check to admin settings
- Add debug statements
- StorageCrawlJob: Restrict search for .ignore files to current storage
v3.2.2
Fixed
- Classifier#getConvertedFilePath: Remove file from queue if it can't be processed
- SettingsService#setSetting: Fix key existence check
v3.2.1
Fixed
- Admin settings: Fix auto-save for batch size settings
v3.2.0
Added
- Add status indicators about Node.js and libtensorflow installation
- Allow setting batch sizes in admin settings
- Support .nomedia tags (thanks to @fa0311)
- DAV faces endpoint: Provide number of files per face
Fixed
- Classifier: Don't process images larger than 8MiB
- Classifier: Use nc preview provider to generate smaller tempfiles
- Don't create duplicate queue entries
- Classifier: Remove file from queue if it can't be found in IRootFolder