* Initial work for adding OpenVino detector. Not functional
* Load model and submit for inference.
Sucessfully load model and initialize OpenVino engine with either CPU or GPU as device.
Does not parse results for objects.
* Detection working with ssdlite_mobilenetv2 FP16 model
* Add OpenVIno support and model to docker image
* Add documentation for OpenVino detector configuration
* Adds support for ARM32/ARM64 and the Myriad X hardware
- Use custom-built openvino wheel for all platforms
- Add libusb build without udev for NCS2 support
* Add documentation around Intel CPU requirements and NCS2 setup
* Print all available output tensors
* Update documentation for config parameters
* Get storage output stats for each camera
* Add storage route
* Add storage route
* Add storage page
* Cleanup
* Add stats and show more storage
* Add tests for mb abbrev util fun
* Rewrite storage logic to use storage maintainer and segment sizes
* Include storage maintainer for http
* Use correct format
* Remove utils
* Fix tests
* Remove total from equation
* Multiply by 100 to get percent
* Add basic storage info
* Fix storage stats
* Fix endpoint and ui
* Fix formatting
* Add hwaccel presets
* Use hwaccel presets
* Add input arg presets
* Use input arg presets
* Make util to clean up redundant code
* Add support for output arg presets
* Add tests
* Update camera specific to use presets
* Update hwaccel to use presets
* Format files and fix tests
* Rewrite tests to test record correctly
* Move presets from string to list to avoid manually separating into a list
* Add mjpeg cuvid decoder preset
* Fix tests
* Fix comment
* Move to images specific folder
* Send error image when camera stream is not available
* Immediately set camera_fps to 0 if camera crashes
* Cache error image so it is not read from file system on each run
* Move camera fps set
* Add ffprobe endpoint
* Get ffprobe for multiple inputs
* Copy ffprobe in output
* Fix bad if statement
* Return full output of ffprobe process
* Return full output of ffprobe process
* Make ffprobe button show dialog with output and option to copy
* Add driver names to consts
* Add driver env var name
* Setup general tracking for GPU stats
* Catch RPi args as well
* Add util to get radeontop results
* Add real amd GPU stats
* Fix missed arg
* pass config
* Use only the values
* Fix vram
* Add nvidia gpu stats
* Use nvidia stats
* Add chart for gpu stats
* Format AMD with space between percent
* Get correct nvidia %
* Start to add support for intel GPU stats
* Block out RPi as util is not currently available
* Formatting
* Fix mypy
* Strip for float conversion
* Strip for float conversion
* Fix percent formatting
* Remove name from gpu map
* Add tests and fix AMD formatting
* Add nvidia gpu stats test
* Formatting
* Add intel_gpu_top for testing
* Formatting
* Handle case where hwaccel is not setup
* Formatting
* Check to remove none
* Don't use set
* Cleanup and fix types
* Handle case where args is list
* Fix mypy
* Cast to str
* Fix type checking
* Return none instead of empty
* Fix organization
* Make keys consistent
* Make gpu match style
* Get support for vainfo
* Add vainfo endpoint
* Set vainfo output in error correctly
* Remove duplicate function
* Fix errors
* Do cpu & gpu work asynchonously
* Fix async
* Fix event loop
* Fix crash
* Fix naming
* Send empty data for gpu if error occurs
* Show error if gpu stats could not be retrieved
* Fix mypy
* Fix test
* Don't use json for vainfo
* Fix cross references
* Strip unicode still
* await vainfo response
* Add gpu deps
* Formatting
* remove comments
* Use empty string
* Add vainfo back in
* Add option for mqtt config
* Setup communication layer
* Have a dispatcher which is responsible for handling and sending messages
* Move mqtt to communication
* Separate ws communications module
* Make ws client conform to communicator
* Cleanup imports
* Migrate to new dispatcher
* Clean up
* Need to set topic prefix
* Remove references to mqtt in dispatcher
* Don't start mqtt until dispatcher is subscribed
* Cleanup
* Shorten package
* Formatting
* Remove unused
* Cleanup
* Rename mqtt to ws on web
* Fix ws mypy
* Fix mypy
* Reformat
* Cleanup if/else chain
* Catch bad set commands
* Refactor mqtt client
* Protect callback method
* Use async to handle reconnects
* Set types and clenup
* Don't set connected until rc code is checked
* Typing: events.py
* Remove unused variable
* Fix return Any from return statement
Not all elements from the event dict are sure to be something that can be evaluated
See e.g.: https://github.com/python/mypy/issues/5697
* Sort out Event disambiguity
There was a name collision of multiprocessing Event type and frigate events
Co-authored-by: Sebastian Englbrecht <sebastian.englbrecht@kabelmail.de>
* Start restream before detection
* Add docs explaining how to reduce connections to the camera
* Fix typos for consistency
* Add link to other part of doc for readability
* Update go2rtc to rc3
* Simplify ffmpeg / audio conversions
* Set ffmpeg bin location
* Manually set video as copied
* Run go2rtc with env vars
* Remove manual ffmpeg declaration
* Enable force_audio by default
* Fix test
* Move each camera to a separate card and show per process info
* Install top
* Add support for cpu usage stats
* Use cpu usage stats in debug
* Increase number of runs to ensure good results
* Add ffprobe endpoint
* Get ffprobe for multiple inputs
* Copy ffprobe in output
* Add fps to camera metrics
* Fix lint errors
* Update stats config
* Add ffmpeg pid
* Use grid display so more cameras can take less vertical space
* Fix hanging characters
* Only show the current detector
* Fix bad if statement
* Return full output of ffprobe process
* Return full output of ffprobe process
* Don't specify rtsp_transport
* Make ffprobe button show dialog with output and option to copy
* Adjust ffprobe api to take paths directly
* Add docs for ffprobe api
* Move existing checks to own functions
* Add config check for zone objects that are not tracked
* Add tests for config error
* Formatting
* Catch case where field is defined multiple times and add test
* Add warning for rtmp
* Refactor EdgeTPU and CPU model handling to detector submodules.
* Fix selecting the correct detection device type from the config
* Remove detector type check when creating ObjectDetectProcess
* Fixes after rebasing to 0.11
* Add init file to detector folder
* Rename to detect_api
Co-authored-by: Nicolas Mowen <nickmowen213@gmail.com>
* Add unit test for LocalObjectDetector class
* Add configuration for model inputs
Support transforming detection regions to RGB or BGR.
Support specifying the input tensor shape. The tensor shape has a standard format ["BHWC"] when handed to the detector, but can be transformed in the detector to match the model shape using the model input_tensor config.
* Add documentation for new model config parameters
* Add input tensor transpose to LocalObjectDetector
* Change the model input tensor config to use an enumeration
* Updates for model config documentation
Co-authored-by: Nicolas Mowen <nickmowen213@gmail.com>
* Consts for regex
* Add regex for camera username and password
* Redact user:pass from ffmpeg logs
* Redact ffmpeg commands
* Move common function to util
* Add tests
* Formatting
* Remove unused imports
* Fix test
* Add port to test
* Support special characters in passwords
* Add tests for special character handling
* Remove docs about not supporting special characters
* add option enabled for each camera in config
* Simplified If-block and removed wrong Optional
* Update Docs enabling/disabling camera in config
* correct format for option
* Disabling Camera for processes, no config changes
* Describe effects of disabled cam in documentation
* change if-logic, obsolete copy, info disabled cam
* changed color to white, added top padding in disabled camera info
* changed indentation
* Pull go2rtc dependency
* Add go2rtc to local services and add to s6
* Add relay controller for go2rtc
* Add restream role
* Add restream role
* Add restream to nginx
* Add camera live source config
* Disable RTMP by default and use restream
* Use go2rtc for camera config
* Fix go2rtc move
* Start restream on frigate start
* Send restream to camera level
* Fix restream
* Make sure jsmpeg works as expected
* Make view rspect live size config
* Tweak player options to fit live view
* Adjust VideoPlayer to accept live option which disables irrelevant controls
* Add multiple options from restream live view
* Add base for webrtc option
* Setup specific restream modules
* Make mp4 the default streaming for now
* Expose 8554 for rtsp relay from go2rtc
* Formatting
* Update docs to suggest new restream method.
* Update docs to reflect restream role
* Update docs to reflect restream role
* Add webrtc player
* Improvements to webRTC
* Support webrtc
* Cleanup
* Adjust rtmp test and add restream test
* Fix tests
* Add restream tests
* Add live view docs and show different options
* Small docs tweak
* Support all stream types
* Update to beta 9 of go2rtc
* Formatting
* Make jsmpeg the default
* Support wss if made from https
* Support wss if made from https
* Use onEffect
* Set url outside onEffect
* Fix passed deps
* Update docs about required host mode
* Try memo instead
* Close websocket on changing camera
* Formatting
* Close pc connection
* Set video source to null on cleanup
* Use full path since go2rtc can't see PATH var
* Adjust audio codec to enable browser audio by default
* Cleanup stream creation
* Add restream tests
* Format tests
* Mock requests
* Adjust paths
* Move stream configs to restream
* Remove live source
* Remove live config
* Use live persistence for which view to use on each camera
* Fix live sizes
* Only use jsmpeg sizes for jsmpeg live
* Set max live size
* Remove access of live config
* Add selector for live view source in web view
* Remove RTMP from default list of roles
* Update docs
* Fix tests
* Fix docs for live view modes
* make default undefined to avoid race condition
* Wait until camera source is loaded to avoid race condition
* Fix tests
* Add config to go2rtc
* Work with config
* Set full path for config
* Set to use stun
* Check for mounted file
* Look for frigate-go2rtc
* Update docs to reflect webRTC configuration.
* Add link to go2rtc config
* Update docs to be more clear
* Update docs to be more clear
* Update format
Co-authored-by: Felipe Santos <felipecassiors@gmail.com>
* Update live docs
* Improve bash startup script
* Add option to force audio compatibility
* Formatting
* Fix mapping
* Fix broken link
* Update go2rtc version
* Get go2rtc webui working
* Add support for mse
* Remove mp4 option
* Undo changes to video player
* Update docs for new live view options
* Make separate path for mse
* Remove unused
* Remove mp4 path
* Try to get go2rtc proxy working
* Try to get go2rtc proxy working
* Remove unused callback
* Allow websocket on restrea dashboard
* Make mse default stream option
* Fix mse sizing
* don't assume roles is defined
* Remove nginx mapping to go2rtc ui
Co-authored-by: Felipe Santos <felipecassiors@gmail.com>
Co-authored-by: Blake Blackshear <blakeb@blakeshome.com>
* Add field and migration for segment size
* Store the segment size in db
* Add comment
* Add default
* Fix size parsing
* Include segment size in recordings endpoint
* Start adding storage maintainer
* Add storage maintainer and calculate average sizes
* Update comment
* Store segment and hour avg sizes per camera
* Formatting
* Keep track of total segment and hour averages
* Remove unused files
* Cleanup 2 hours of recordings at a time
* Formatting
* Fix bug
* Round segment size
* Cleanup some comments
* Handle case where segments are not deleted on initial run or is only retained segments
* Improve cleanup log
* Formatting
* Fix typo and improve logging
* Catch case where no recordings exist for camera
* Specifically define sort
* Handle edge case for cameras that only record part time
* Increase definition of part time recorder
* Remove warning about not supported storage based retention
* Add note about storage based retention to recording docs
* Add tests for storage maintenance calculation and cleanup
* Format tests
* Don't run for a camera with no recording segments
* Get size of file from cache
* Rework camera stats to be more efficient
* Remove total and other inefficencies
* Rewrite storage cleanup logic to be much more efficient
* Fix existing tests
* Fix bugs from tests
* Add another test
* Improve logging
* Formatting
* Set back correct loop time
* Update name
* Update comment
* Only include segments that have a nonzero size
* Catch case where camera has 0 nonzero segment durations
* Add test to cover zero bandwidth migration case
* Fix test
* Incorrect boolean logic
* Formatting
* Explicity re-define iterator
* Set sub label on object data if event is in progress
* Include sub_label in dict
* Don't need to set and passively get
* Formatting
* Don't expect event to be valid
* Update docs to reflect that sub label is included
* -1 so ensure indexes are correct
* Catch case of zero division
* Due to the -1, it may be negative
* Ignore source of error
The error is occurring due to a detections bounding box starting beyond the frame, this should be immediately ignored
* Formatting
* Check horizontal placement as well
* Remove original frame clamping
* Fix VOD issues with longer keyframe intervals
* Move probe function to util
Update comment
* Use recording duration for keyFrameDurations
* Remove unused early return
* Avoid clipping first clip
* Add log message when discarding recording segments in cache
Currently Frigate silently discards recording segments in cache if there's more than "keep_count" for a camera, which can happen on high load/busy/slow systems.
This results in recording segments being lost with no apparent cause in the logs (even when set to debug).
This PR adds a warning log entry when discarding segments, in the same way as discarding corrupted segments
* Add explanatory warning and properly format cache_path warning
* lint fixes
Co-authored-by: Blake Blackshear <blakeb@blakeshome.com>
* Set up for http tests
* Setup basics for testing and first test
* Add testing consts
* Cleanup db creation
* Add one more check to test
* Get event that does not exist
* Get events working with cleaner db
* Test retain / un-retain
* Test setting and deleting sub label
* Test getting list of sub labels
* Fix bug caught in tests
* Test deleting event
* Test geting list of events
* Expand test
* Test more event filters
* Write version module so tests don't fail on version import
* Test config
* Test recordings endpoint
* Formatting
* Remove unused imports
* Test stats
* Add cleanup files in const
* Add name to match other checks
* Use default names so filters are more clear
* Add endpoint to get list of sub labels inside DB
* Fix crash on no internet
* Cleanups for sub_label http
* Add sub label selector to events UI
* Add event filtering for sub label
* Formatting files
* Reduce size of filters to fit on one line
* Add handler for tests
* Remove unused imports
* Only show the sub labels filter when there are sub labels in the DB
* Fix tests
* Use distinct instead of group_by
* Formatting
* Cleanup event logic
* Send mqtt message when motion is detected
* Use object processing instead of passing mqtt client around
* Cleanup
* Formatting
* add comment
* Make off delay configurable.
* Handle updating each camera based on config off delay
* Formatting
* Update docker-compose.yml
* Fix processing issue
* Update mqtt docs
* Update main config docs
* Make sure multiple True values aren't published for the same motion
* Make sure multiple True values aren't published for the same motion
* Update payload to fit existing HA standard values
* Update docs to fit new values
* Update docs
* Update motion topic
* Use datetime.datetime and remove unused imports
* Cast to int
* Clarify motion detector behavior in docs
* Fix typo
Co-authored-by: Blake Blackshear <blakeb@blakeshome.com>
* Starting working on adding motion toggle
* Add all info to mqtt command
* Send motion to correct funs
* Update mqtt docs
* Fixes for contingencies
* format
* mypy
* Tweak behavior
* Fix motion breaking frames
* Fix bad logic in detect set
* Always set value for motion boxes
* __main__.py
* app.py
* models.py
* plus.py
* stats.py
In addition a new module was introduced: types
There all TypedDicts are included. Bitte geben Sie eine Commit-Beschreibung für Ihre Änderungen ein. Zeilen,
* Toggle improve_contrast for cameras via MQTT
* Process parameter to mqtt toggle improve_contrast
* Update mqtt docs for improve_contrast topic
* Spacing
* Add class variable and update in process_frames
* Pass to constructor
* pass by reference mistake
* remove parameter
* remove parameter
* Add options for reordering and hiding cameras selectively
* Add newline at end of camera file
* Make each camera for birdseye togglable as well
* Update names to be less ambiguous
* Update defaults
* Include sidebar change
* Remove birdseye toggle (will be added in separate PR)
* Remove birdseye toggle (will be added in separate PR)
* Remove birdseye toggle (will be added in separate PR)
* Update sidebar to only sort cameras once
* Simplify sorting logic
* Add camera level processing for birdseye
* Add camera level birdseye configruation
* Propogate birdseye from global
* Update docs to show that birdseye is overridable
* Fix incorrect default factory
* Update note to indicate values that can be overridden
* Cleanup config accessing
* Add tests for birdseye config behavior
* Fix mistake on test format
* Update tests
* Add object ratio config parameters
Issue: #2948
* Add config test for object filter ratios
Issue: #2948
* Address review comments
- Accept `ratio` default
- Rename `bounds` to `box` for consistency
- Add migration for new field
Issue: #2948
* Fix logical errors
- field migrations require default values
- `clipped` referenced the wrong index for region, since it shifted
- missed an inclusion of `ratio` for detections in `process_frames`
- revert naming `o[2]` as `box` since it is out of scope!
This has now been test-run against a video, so I believe the kinks are
worked out.
Issue: #2948
* Update contributing notes for `make`
Issue: #2948
* Fix migration
- Ensure that defaults match between Event and migration script
- Deconflict migration script number (from rebase)
Issue: #2948
* Filter objects out of ratio bounds
Issue: #2948
* Update migration file to 009
Issue: #2948
* Add sub label to model and set / delete funs
* Add migrations for sub label
* Tweaks to API and model
* Show sublabel if available
* Cleanups
* Update docs
* Show person in UI title
* Fix typo and don't fail on no json
* Transfer sub labels for in progress events
* Remove sublabel reset
* Remove person only check
* Make default null
* Update docs and formatting
* Make default null
* Make nullable in migration
* Undo null
* Update model to accept null
* Update migration to accept null
* Don't set to default values
* Remove redundant defaults and update http logic
* Only need a single route
* Enforce 20 character limit in http
* Update docs to mention 20 character limit
* Cleanup
* Separate insert and update to make sure updated values are retained when event ends
* Use insert instead of replace
* Remove redundant if and have should_update_db include clip or snapshot requirement.
* Added object thumbnail def and made camera tracked objects use it.
* Add object snapshot def
* Remove documentation for best.jpg
* Update docs for label thumbnail and snapshot defs
* 📝✅🔧 - Make RTMP config global
Fixes#1671
* 📝✅🔧 - Make timestamp style config global
Fixes#1656
* fix test function names
* formatter
Co-authored-by: Blake Blackshear <blakeb@blakeshome.com>
If jpg_bytes wasn't retrieved from either desk or a tracked object, respond with 404
Prevents uncaught error for unknown event ids sent to event_snapshot endpoint
Lock updates to tracked objects, current frame time, motion boxes, and
regions on `update()`.
Directly create Counters using counted values.
Don't convert removed_ids, new_ids, or updated_ids sets to lists.
Update defaultdict's to remove un-necessary lambdas when possible.
When possible, drop un-necessay list comprehensions, such as when
calling `any`.
Use set comprehension, rather than passing a list comprehension into
`set()`.
Do the slightly more pythonic `x not in y` rather than `not x in y` to
check list inclusion.
Use config data classes to eliminate some of the boilerplate associated
with setting up the configuration. In particular, using dataclasses
removes a lot of the boilerplate around assigning properties to the
object and allows these to be easily immutable by freezing them. In the
case of simple, non-nested dataclasses, this also provides more
convenient `asdict` helpers.
To set this up, where previously the objects would be parsed from the
config via the `__init__` method, create a `build` classmethod that does
this and calls the dataclass initializer.
Some of the objects are mutated at runtime, in particular some of the
zones are mutated to set the color (this might be able to be refactored
out) and some of the camera functionality can be enabled/disabled. Some
of the configs with `enabled` properties don't seem to have mqtt hooks
to be able to toggle this, in particular, the clips, snapshots, and
detect can be toggled but rtmp and record configs do not, but all of
these configs are still not frozen in case there is some other
functionality I am missing.
There are a couple other minor fixes here, one that was introduced
by me recently where `max_seconds` was not defined, the other to
properly `get()` the message payload when handling publishing mqtt
messages sent via websocket.
Use `np.unique` to determine the correct set of row/col pairs to iterate
over when doing the object matching without needing to track which rows
or columns have already been seen. Add to some of the accompanying
documentation to clarify this algorithm.
Also fix what looks to be an erroneous early return, and change this to
a continue.
Generally eliminate the `while True` loops while waiting for a stop
event and prefer to condition the loops on if the stop event is set,
blocking on that where it makes sense. This generally comes in 3
flavors. First and simplest, when there is a sleep and the stop event
is the only thing the loop blocks on, instead do a check using
`stop_event.wait(timeout)` to instead block on the stop event for the
designated amount of time. Second, when there is a different event that
is blocking in the loop, condition the loop on `stop_event.is_set()`
rather than breaking when it is set. Finally, when there is a separate
internal condition that requires a counter, have the loop iterate over
the counter and use `if stop_event.wait(timeout)` internal to the loop.
When running ffprobe, use `subprocess.run` rather than
`subprocess.Popen`. This simplifies the handling that is needed to run
and process the outputs. Here, filename parsing is also simplified by
explicitly removing the file extension with `os.path.splitext` and
forcing a single split into the camera name and the formatted date.
StatsEmitter thread to send stats to MQTT every 60 seconds by default, optional stats_interval config value.
New service stats attribute, containing uptime in seconds and version.
