Viseron

This release introduces config reloading without restarting, external motion detector support, a Home Assistant App version, and upgrades to Ubuntu 24.04, PostgreSQL 16, and FFmpeg 8.1.

This release fixes player control visibility on hybrid devices and improves the Azure build process.

Re-release of v3.5.1 to enable builds using self-hosted amd64 agents in Azure pipelines to resolve memory issues.

Fixed a bug where certain window heights caused timeline clicks to misinterpret 10-digit timestamps with decimals as millisecond timestamps.

This release fixes Hailo component configuration referencing and sets the multi_process_service default to false.

This release introduces a complete frontend redesign, reverse proxy subpath support, public image URLs for notifications, and various performance improvements including a React 19 upgrade.

Fixed the Hailo-8 default model download URL.

Introduces webhook and Hailo-8 components, a system event viewer, a template editor, and timelapse functionality in the storage component.

This release fixes a race condition during camera restarts and resolves issues where the retry loop could get stuck after a camera disconnection.

Fixed a race condition and a retry loop issue that caused cameras to remain disconnected when rapidly starting and stopping.

Fixes an uncaught exception that previously caused object detection to stop working in certain scenarios.

Introduces a new Live page and go2rtc component for improved WebRTC/MSE streaming, adds a YOLOv8 object detection component, and enhances EdgeTPU and tiered storage performance.

Fixes a bug that prevented EdgeTPU image classification from functioning.

Fixes a bug that required both recorder and snapshots to be specified together in the camera storage configuration.

Introduces multi-user support with role-based access control, user management, and camera assignment, alongside new notification components and post-processor masks.

Introduces 24/7 continuous recordings, storage tiers, a new timeline and events view, a safe mode for config errors, and various dependency updates.

This beta release introduces stability improvements, new Telegram and PTZ components, improved event downloads, and updated documentation, while renaming the recordings folder to event_clips.

This release improves stability through numerous bug fixes and introduces scheduled cleanup jobs to synchronize the database and filesystem.

This release introduces player synchronization for Event/Timeline views, significant performance optimizations for the event table, and upgrades to React 18, MUI v6, and React Query v5.

Introduces a new multi-camera grid layout for the Events page, event filtering, HLS streaming improvements, and various performance and stability fixes.

Fixed a crash issue related to edgetpu.

Updates OpenCV to 4.10.0, improves image annotations, fixes HLS playback, and enhances snapshot storage and event visibility for license plate and motion detection.

Added Face Recognition events, grouped events tab, improved log redaction, and a new storage check_interval config option.

Fixed GStreamer pipeline hangs on Ubuntu 22.04 and added support for VAAPI decoding.

Fixed file deletion logic and improved cross-build compatibility for gpac on aarch64 and Jetson Nano, including PostgreSQL version detection.

This release updates containers to Ubuntu Jammy and Python 3.10, upgrades several core libraries including CUDA, OpenCV, and Dlib, and improves PostgreSQL upgrade handling and UI elements.

Fixed failing Azure builds with a minor version update.

This release includes various fixes for thumbnails, HLS playback, the Events page layout, and database connection permissions.

Fixed a missing 'gi' module dependency for the jetson-nano container to resolve GStreamer loading failures.

Introduces a CodeProject AI component with ALPR support, enhances developer experience with VSCode Devcontainers and documentation, and adds new configuration options for still images and FFmpeg raw commands.

Introduces optional frontend and API authentication, improved camera retry logic, and various FFmpeg and library updates.

Introduces optional frontend/API authentication, improved camera setup retries, FFmpeg 5.1.2 upgrade, and various bug fixes for audio transcoding and hardware acceleration.

This minor release introduces recording deletion via UI, customizable ffmpeg output arguments, and aarch64 docker image updates, while fixing a critical recording length bug.

Introduces daily recording grouping, breadcrumbs, OpenCV 4.6.0 with OpenVINO support, and makes YOLOv7 the default.

Viseron v2 introduces a modular rewrite with a new configuration format, a React-based UI, improved Jetson Nano support, and new face recognition and image classification capabilities.

Introduces a new Entities Page for browsing entity states, adds YOLOv7 models, and includes various performance improvements and bug fixes.

This release fixes bounding box and permission issues, upgrades dlib to 19.24 for better CUDA support, and includes various documentation and scaling improvements.

Viseron v2 introduces a modular rewrite with a completely new configuration format, a new React-based UI, improved Jetson Nano support, and a dedicated documentation site.

Fixed a deadlock issue that occurred when using a substream.

Bugfix release that properly overrides configuration for motion detection.

Introduces a basic UI, new motion detectors, expanded configuration options for EdgeTPU and object detection, and CUDA updates.

This release adds DeepStack object and face detection, Jetson Nano support, and improved stability and camera reconnection handling.

This release revamps Docker containers with multistage builds, introduces experimental RPi4 support, adds static MJPEG streams, and implements new configuration options for labels and face recognition.

Introduces a new save_unknown_faces config for dlib, experimental default audio recording, and switches the default model to YOLOv4.

Introduces a require_motion config option for labels, fixes label corruption in amd64 images, and updates Docker builds for reduced size and multiarch support.

Re-release of 1.8.0b3 featuring rewritten Dockerfiles for smaller sizes, multi-arch support, static MJPEG streams, and non-root container execution.

This release contains no changes; please refer to the next release.

Fixed a bug to restore functionality to filter_args.

Added MJPEG streams for cameras with configurable query parameters and improved EdgeTPU error reporting.

Fixed an issue where the system would attempt to publish to MQTT without checking if it was enabled.

Fixed an issue to ensure thumbnails are correctly saved to disk.

Introduces FFmpeg segments for recording, substream support for image processing, CUDA/OpenCV upgrades, and improved startup speed via parallel camera setup.

Introduces FFmpeg segments and substreams for improved recording and image processing, updates CUDA and OpenCV, and optimizes recording storage paths.

Fixes a crash occurring when MQTT is not connected.

This release includes fixes for issues #60 and #61 and provides updated Docker images.

Introduces face detection and improved Home Assistant discovery, while updating MQTT topics and configuration validation for detectors.

Adds PCIe EdgeTPU support, a new object detection logging configuration, improved codec picking, and uses ffprobe for stream information.

Introduces face detection, improves Home Assistant discovery, and updates default YOLO models, alongside significant changes to MQTT topics and configuration validation.

Introduces motion detection masks, Home Assistant arm/disarm switches, and new configuration options for motion timeouts and RTSP transport.

Introduces camera zones for targeted object tracking, expanded binary sensors, granular logging controls, and individual camera configuration for dimensions and FPS.

Introduces per-camera logging, a new publish_image config option, individual camera resolution and FPS settings, and fixes for zone-based binary sensors.

Introduces camera zones allowing users to track specific labels within defined polygon areas and create corresponding binary sensors.

Added support for mjpeg streams and fixed issues with detector intervals, memory usage for zero lookback, and codec overwriting.

Introduces secrets.yaml for config value substitution, adds benchmarks to the README, and fixes a decoder crash related to default motion detection.

Updates object detection configuration to be per-label, adds per-camera detection settings, refactors Darknet for performance, and improves logging and codec configuration.

Initial release of Viseron with Docker images available for various platforms.