0.7.0 Release Candidate 1

Changes:

• Raspberry Pi 3/4 Support
• Update ffmpeg to 4.3.1
• Support multiple coral devices
• Reduce CPU usage of ffmpeg and motion detection
• Replace underlying memory store with shared_memory and update to Python 3.8
• Smaller docker images due to switch to multistage builds
• shm-size parameter no longer necessary for the majority of scenarios

Breaking Changes:

• MQTT snapshots no longer draw regions or bounding boxes
• Default output_args for cameras have changed. If you specified custom output parameters, you will need to update.
• Updated config for specifying coral device. See new detectors section in the example config.
• New json structure at /debug/stats to handle multiple detectors

Images:

• x86: blakeblackshear/frigate:0.7.0-rc1
• Raspberry Pi 3/4 (32bit): blakeblackshear/frigate:0.7.0-rc1-armv7hf
• Raspberry Pi 4 (64bit): blakeblackshear/frigate:0.7.0-rc1-arm64

0.6.1 Release

• Remove car and truck from default list
• Replace configs instead of merge to avoid confusion
• Fix zone filter issue #218

0.6.0 - Zones, Clips, and Less False Positives

Changes:

• Ability to save a video clip of each tracked object
• New event mqtt topics
• An updated snapshot is now sent via MQTT when a higher scoring object is detected
• Ability to define a zone to detect objects in specific parts of the image
• Mask can now be specified as base64 in the config file for easier k8s config
• Frigate now exits gracefully for SIGINT and SIGTERM
• New latest.jpg endpoint for fetching the current frame
• Best images can now be resized with a query string parameter to grab lower resolution images for notifications
• Updated method for computing a median score that reduces false positives
• New min_score and threshold definitions
• Allows configuring a specific EdgeTPU device for the container
• Modified labelmap to treat trucks as car to avoid confusion between the two
• Crop and resize the best.jpg endpoint
• Made best.jpg image timeout configurable
• Crop and resize images before sending to MQTT
• Limit what object types trigger recordings
• Limit cache size
• Polygon mask type
• Added affiliate links to the readme

Breaking Changes:

• In previous versions, ffmpeg used -vsync drop as an input parameter. It is incompatible with the new save_clips feature, so you will need to remove it if you added it directly to your config.

Image available at docker pull blakeblackshear/frigate:0.6.0.

0.6.0 Release Candidate 3 - Zones, Clips, and Less False Positives

• Made best.jpg image timeout configurable

0.6.0 Release Candidate 2 - Zones, Clips, and Less False Positives

• Updated method for computing a median score that reduces false positives
• New min_score and threshold definitions
• Allows configuring a specific EdgeTPU device for the container
• Modified labelmap to treat trucks as car to avoid confusion between the two
• Crop and resize the best.jpg endpoint
• Crop and resize images before sending to MQTT
• Limit what object types trigger recordings
• Limit cache size
• Polygon mask type
• Added affiliate links to the readme

0.6.0 Release Candidate 1 - Zones and Clips

Changes:

• Ability to save a video clip of each tracked object
• New event mqtt topics
• An updated snapshot is now sent via MQTT when a higher scoring object is detected
• Ability to define a zone to detect objects in specific parts of the image
• Mask can now be specified as base64 in the config file for easier k8s config
• Frigate now exits gracefully for SIGINT and SIGTERM
• New latest.jpg endpoint for fetching the current frame
• Best images can now be resized with a query string parameter to grab lower resolution images for notifications

Breaking Changes:

• In previous versions, ffmpeg used -vsync drop as an input parameter. It is incompatible with the new save_clips feature, so you will need to remove it if you added it directly to your config.

Image available at docker pull blakeblackshear/frigate:0.6.0-rc1.

0.5.2 Release

• Fixes bug that causes stationary objects to be lost sometimes
• Fix model path from Google repos
• Switch to MIT License

Available at blakeblackshear/frigate:0.5.2' and blakeblackshear/frigate:stable'

0.5.1 Release

Fixes for several different hanging scenarios.

Changes:

• Added tzdata to image to allow setting timezone with TZ env var
• Removed unused SharedArray python package
• Remove fps config value for setting expected framerate of the camera
• Skip frames in capture thread when needed to avoid filling plasma store
• Handle plasma store failures
• Fix race condition on detection process watchdog checks
• Move ffmpeg process to separate thread and communicate with a queue
• Ensure frames are cleaned up from plasma store
• Resize debug view based on actual aspect ratio of the frame
• Add frame_time for each step in processing in /debug/stats

Docker image is available with docker pull blakeblackshear/frigate:0.5.1

0.5.1 Release Candidate 4

• Resize preview according to camera's aspect ratio
• Discard plasma store logs
• Fix check for detection process hanging

0.5.1 Release Candidate 3

• Improve cleanup of frame cache in plasma store
• Add some debugging to see which frame each process is working on
• Skip frames in the capture process if the detection process is falling behind