mirror of
https://github.com/blakeblackshear/frigate.git
synced 2024-12-29 00:06:19 +01:00
NVR with realtime local object detection for IP cameras
aicameragoogle-coralhome-assistanthome-automationhomeautomationmqttnvrobject-detectionrealtimertsptensorflow
9df5927ac5
* zoom in/out in search for lost objects * predicted box should not be empty * clean up and update zoom logic * only zoom if enabled * more cleanup * check for valid velocity when zooming * only try absolute zoom in if obj area has changed * zoom logic * don't enqueue lost object zoom if already at limit * don't disable motion boxes during ptz moves * velocity threshold based on move coefficients * fix area zoom logic * disable debug zoom * don't process objects if ptz moving * recalc with exponent * change exponent * remove lost object zooming * increase distance threshold for stationary object * increase distance threshold constant * only zoom out if nonzero * camera name in all debug logging * add camera name to debug logging * camera variable name consistency * update calibration behavior and docs * docs and better zooming * more sensible target values * docs wording * fix velocity threshold variable * zooming tweaks and remove iou for current objects * debug and docs * get valid velocity * include zero * additional debug statements * add zoom hysteresis * zoom on initial move if relative * only update target box if we actually zoom * merge dev * use getattr instead of get * increase distance threshold * reverse logic * get_camera_status after preset move to store zoom * final tweaks and docs * use constants and catch possible debug exception * adjust zoom factor exponent * don't run motion estimation when calling preset * adjust dimension threshold * use numpy for velocity estimate calcs * more numpy conversion * fix numpy shapes * numpy zeros dimension * more zoom out conditions * fix velocity bug * ensure init has been called in debug view * ensure onvif init if enabling by mqtt * change default hysteresis values * recalc relative zoom value * zoom out value * try to zoom when object isn't moving * try zoom when tracked object is not moving * don't try to zoom every time * negate zoom out condition when needed * hysteresis constants for absolute zooming * update zoom conditions * don't recalc target box on zoom only * zoom out if above area threshold * don't print zooming debug for stationary obj * revamp zooming to use area moving average * zooming tweaks and expose property * limit zoom with max target box * use calibration to determine zoom levels * zoom logic fix * docs * add tapo c200 camera * fix initial absolute zoom * small zoom logic fix * better invalid velocity checks * fix test * really fix test this time |
||
|---|---|---|
| .devcontainer | ||
| .github | ||
| .vscode | ||
| config | ||
| docker | ||
| docs | ||
| frigate | ||
| migrations | ||
| web | ||
| .dockerignore | ||
| .gitignore | ||
| .pylintrc | ||
| audio-labelmap.txt | ||
| benchmark_motion.py | ||
| benchmark.py | ||
| CODEOWNERS | ||
| docker-compose.yml | ||
| labelmap.txt | ||
| LICENSE | ||
| Makefile | ||
| process_clip.py | ||
| pyproject.toml | ||
| README.md | ||
Frigate - NVR With Realtime Object Detection for IP Cameras
A complete and local NVR designed for Home Assistant with AI object detection. Uses OpenCV and Tensorflow to perform realtime object detection locally for IP cameras.
Use of a Google Coral Accelerator is optional, but highly recommended. The Coral will outperform even the best CPUs and can process 100+ FPS with very little overhead.
- Tight integration with Home Assistant via a custom component
- Designed to minimize resource use and maximize performance by only looking for objects when and where it is necessary
- Leverages multiprocessing heavily with an emphasis on realtime over processing every frame
- Uses a very low overhead motion detection to determine where to run object detection
- Object detection with TensorFlow runs in separate processes for maximum FPS
- Communicates over MQTT for easy integration into other systems
- Records video with retention settings based on detected objects
- 24/7 recording
- Re-streaming via RTSP to reduce the number of connections to your camera
- WebRTC & MSE support for low-latency live view
Documentation
View the documentation at https://docs.frigate.video
Donations
If you would like to make a donation to support development, please use Github Sponsors.
Screenshots
Integration into Home Assistant
Also comes with a builtin UI:
