Frigate

Maintenance release addressing multiple security vulnerabilities, improving live page preview performance, and adding optional recording segment chapters to mp4 exports.

Maintenance release 0.17.1 featuring security fixes for config endpoints and camera authentication, alongside various bugfixes and documentation updates.

Version 0.17.0 introduces local classification model training, GenAI review summaries, semantic search triggers, local audio transcription, and significant hardware acceleration improvements for Nvidia, Intel, RKNN, and Apple Silicon.

This release addresses a security vulnerability regarding authenticated remote command execution and container escape, alongside various documentation updates and a go2rtc check port.

This release addresses a critical security vulnerability regarding broken access control, improves recording playback efficiency, and includes various documentation updates and bug fixes.

Release 0.16.2 addresses a security vulnerability regarding arbitrary file reads, improves UI live loading, and includes various bug fixes and minor feature enhancements.

Maintenance release 0.16.1 providing various frontend and backend bugfixes, including recording playback improvements, autotracking stability, and hardware acceleration fixes.

Version 0.16.0 introduces face and license plate recognition, a new Viewer role, expanded hardware support for Hailo8 and ONNX, and significant UI improvements including i18n support.

This maintenance release for Frigate 0.15 fixes 'too many open files' errors on HAOS 16 by updating ulimit settings.

Maintenance release 0.15.1 fixing a crash in get_current_frame() and the TRT model prepare script.

Version 0.15.0 introduces a new Explore pane with AI-powered semantic and generative search, native WebPush notifications, support for ONNX and Hailo-8 detectors, and significant backend updates including ffmpeg 7.0.

This maintenance release for Frigate 0.14 provides various bug fixes, performance improvements, and UI enhancements across the recordings, review, and live player views.

Version 0.14.0 introduces a completely rewritten UI with a new dashboard and review system, adds user authentication and TLS support, and includes various hardware acceleration improvements.

This release includes improvements to ONVIF handling, birdseye enhancements, NMS optimization for stationary cars, and various documentation updates.

This release includes documentation fixes and an increase in hash map size.

Version 0.13.0 introduces audio events, PTZ autotracking, a new object tracker, and improved recording management, alongside critical security fixes and expanded hardware support.

This release reduces CPU usage for Raspberry Pi users with hwaccel, fixes recording and UI issues, and provides extensive documentation updates.

up’s--- C-S-T-I-S-T-A-S-C-A-C-T-S-T-O-T-A-L-S-T-R-A-T-E-G-Y-S-T-R-A-T-E-C-T-O-R-S-S-T-O-T-AHFrSاتے-Cي-S-T-R-_S-T-E-P-S-T-O-T-A- la-S-T-R-A-T-E-G-Y-S-T-R-A-C-T-OL__:C-O-N-S-T-R-U-C-T-O-R-S-S-T-O-T-S-S-T-R-A-T-E-G-Y-S-T-R-A-T-E-G-Y-S-T-R-A-C-T-O-R-S-S-T-O-T-A-L-S-T-R-A-T-E-G-Y-S-T-R-A-T-E-G-Y-S-T-R-A-T-E-G-Y-S-T-R-C-T-O-R-S-T-O-T-A-L-S-T-S-T-R-A-T-E-G-Y-S-T-R-A-T-E-G-Y-S-T-R-A-T-E-G-Y-S-T-R-A-T-E-G-Y-S-T-R-A-T-E-G-Y-S-T-R-A-T-E-G-Y-S-T-R-A-T-E-G-Y-S-T-R-A-T-E-G-Y-S-T-R-A-T-E-G-Y-S-T-R-A-T-E-G-Y-S-T-R-A-T-E-G-Y-S-T-R-A-

This release introduces MQTT sub labels, various bug fixes for segments and ffprobe, and updates to the camera documentation.

Version 0.11.0 introduces a revamped Events and Recordings view, Frigate+ integration, a multiarch Docker build, and various MQTT enhancements.

Dynamic contrast is now optional and disabled by default, with additional print statement cleanup.

This release fixes an issue with the RTMP global configuration.

This release includes performance improvements to the cache loop and several bug fixes, including reverts to default segment formats and audio inclusion.

Version 0.9.2 includes bug fixes for recording expiration and enhancements to the recording cache, log messaging, live view sizing, and default audio settings.

This release merges clips into recordings, moves detect settings in config, and introduces a recordings viewer, Birdseye view, and JSMpeg live view.

This release introduces a recordings viewer, Birdseye grid view, and optimized live view, while merging clips into recordings and restructuring detection configuration.

Fixed mjpeg stream blocking on the web server and improved WebUI scrolling and stability.

Updated baseUrl handling for addon ingress and websockets.

This release introduces frontend improvements, disk usage stats, global object masks, and proactive cache cleanup.

This release improves FFmpeg log capturing, optimizes WebUI performance, and fixes Firefox live view and event page functionality.

Version 0.8.0 introduces 24/7 recording, a new HomeAssistant integration, a basic web UI, and significant updates to stream handling and object detection.

Implements responsive image sizing for better performance on small screens and fixes relative paths for ingress usage.

Introduces SqliteQueueDatabase to resolve database locking issues in systems with numerous cameras.

This release improves cache handling, adds a snapshot endpoint, fixes thumbnail issues, implements rate limiting for event notifications, and updates documentation.

This release includes fixes for addon ingress, MQTT switch initial states, and frame cache edge cases.

This release includes fixes for ingress, UI colors, MQTT handling, boot detection, and numpy pinning for Jetson Nano.

Introduces a web UI, expanded MQTT functionality, object detection masks, and updated configuration options.

This release includes bug fixes for builtin tmpfs support, mask error handling, and recordings expiration, along with an updated readme.

This release updates the detection model to mobiledet, introduces configurable model dimensions and motion detection, and enhances MJPEG and thumbnail crop functionality.

This beta release introduces a new HomeAssistant integration, 24/7 recording, multiple stream roles per camera, and a revamped MQTT event system.

Fixes bugs related to cache handling and fps tracking introduced in version 0.7.2.

This update improves frame management, moves ffmpeg capture to a separate process, fixes image channel errors, and adds VAAPI support for 10th gen Intel processors.

{"L{"description":"Fixed a bug related to cache and clips directories following the addition of configuration options.","breaking_changes":false}

Adds Raspberry Pi 3/4 support, multiple Coral device support, and performance improvements with Python 3.8 and updated ffmpeg.

Fixed zone status bouncing and added timestamps and optional bounding boxes to snapshot images.

Fixed an issue where the camera process would hang during a detection process restart.

Adds Raspberry Pi 3/4 support, multiple coral device support, and performance optimizations while updating ffmpeg and Python versions.

Removed car and truck from default list, switched to config replacement instead of merging, and fixed a zone filter issue.

Introduces video clip saving, zone detection, base64 masks, and various MQTT and image processing enhancements.

Added a configuration option for the best.jpg image timeout.

Updates include improved median score computing, new configuration options for EdgeTPU and scoring, image preprocessing for MQTT and endpoints, and refined object detection logic.

Adds video clip saving, zoning, new MQTT topics, base64 mask support, and a latest.jpg endpoint.

This release fixes a stationary object bug, corrects the Google model path, and switches to the MIT License.

This release addresses multiple hanging scenarios, improves frame handling and plasma store management, and updates Docker image dependencies.

This update improves camera preview aspect ratio handling, removes plasma store logs, and fixes a detection process hanging issue.

Improved plasma store frame cache cleanup, added process debugging, and implemented frame skipping when detection lags.

Moved ffmpeg collection to a separate thread using a queue for improved processing.

Updated the detection queue for resilience, improved plasma store handling, and replaced Queue with SimpleQueue, removing SharedArray as a dependency.

Major overhaul introducing multi-process architecture for improved performance, optional Coral usage, and lightweight motion detection.

Improved camera recovery after detection hangs and updated the TensorFlow Lite wheel URL.

Improved error handling in the detection process and ensured the restart function is called when detection hangs.

Improved stability by handling ffmpeg restarts, preventing detection process hangs, implementing a queue system, and adding MQTT environment variable support.

Improved ffmpeg subprocess handling by increasing buffer size, modifying process termination logic, and enhancing logging.

This release includes a dictionary iteration fix, enhanced plasma store logging, improved memory cleanup for frames, and additional statistics metrics.

Fixes watchdog calculation and object expiration issues, and restores the ability to specify frame size in the configuration.

Major overhaul introducing multi-process architecture for improved performance, optional Coral support, and lightweight motion detection.

This update introduces object tracking, dynamic regions, a new debug stats endpoint, and UDP camera feed support, while refining detection confidence and image size.

Frigate 0.3.0 introduces per-object configuration, color-coded bounding boxes, and optimized MJPEG feeds.

This release introduces configuration improvements, a health check endpoint, updated ffmpeg args, and person area filtering.

This release introduces significant configuration changes, new person area filtering, custom MQTT client IDs, a health check endpoint, and revamped ffmpeg arguments.

Improved Home Assistant integration via MQTT, enhanced ffmpeg process management, and added configurable ffmpeg arguments.

Introduces hardware accelerated video decoding via FFMPEG, reduces Docker image size, and adds new frame rate options and labeling improvements.

This beta release introduces hardware accelerated video decoding, a new frame rate reduction option, improved label positioning, and a significantly smaller Docker image size.

Introduces configurable regional thresholds, an RTSP stream failure watchdog, and fixes a missing numpy import for mask files.

Adds person detection masking support using BMP files to ignore detections based on black pixels.

Added Google Coral and multi-camera support, switched to config files, bundled detection models, and replaced motion detection with optimized object detection and simplified MQTT messaging.