Viseron

Breaking changes

• Recorder global_args can no longer be specified

Changes and new Features

• FFmpeg segments are now used to record instead of caching frames in memory
• A substream can now be configured which will be used for image processing. This can be used to great success to reduce system load.
• Upgrades to CUDA 11 and OpenCV 4.5.0 for CUDA image
• Recordings are encoded to /tmp, then moved to target directory when done. Closes #49
• Installs tzdata package. Closes #55
• A new config option, thumbnail is available under recorder. Use this to send a snapshot of the detection that triggered a recording to MQTT, or choose to save to disk
• A new sensor entity is created in Home Assistant which indicates the status of Viseron.
  The state is set to recording, scanning_for_motion or scanning_for_objects
• Cameras are now setup in parallel instead of sequentially. This significantly increases startup speed with many cameras

Fixes

• Fixed a formatting issue with the duplicate log filter
• Returncodes > 0 from MQTT is now logged as errors in text format. Closes #59
• Fixes segments process not starting when using substream

Docker images are available on Docker Hub roflcoopter/viseron:1.7.0 roflcoopter/viseron-cuda:1.7.0 roflcoopter/viseron-vaapi:1.7.0 roflcoopter/viseron-rpi:1.7.0