0.17.0 Release #22137
Replies: 29 comments 83 replies
|
Congrats on the release! Lots of great new features—classification training, custom viewer roles, and Apple Silicon support are exciting. Thanks for your hard work! \ud83c\udf89 |
|
Looking forward to what's to come. I appreciate all of your hard work! Thank you |
|
Amazing improvements and great update. Thanks to everyone involved for this wonderful release. |
|
Amazing release ! I'm going to update as soon as available on HA ! |
|
Finally!! Thx for the Amazing update! I always waited for a feature like Classification Model Training, now I can train it on only my cats :D will keep my Frigate Subcription |
|
It looks like the state classification example in the doc duplicated the gate closed image rather than show both open and closed images. But wow, what an incredible update! |
|
I haven't followed the 0.17 development. Are there any specifics that need addressing before updating from 0.16.4? |
|
Then you are ready to go when config and database are backed up. I myself am also HA, no problems til now |
|
Upated from 16.4 five days ago. Not using any new features. |
|
Can I define the crop area for state classification myself? When selecting from the GUI, I cannot define a rectangle, only a square. This makes the area unnecessarily large. |
|
It needs to be a square, it does not matter if there is parts of the area that are not related to the state |
Why do i get "alerts" for what should be "detections" ? |
|
The message under the Alerts checkboxes tells you why: All Person and Car objects on Einfahrt will be shown as Alerts This is Frigate's default behavior. If you want to change it, you'll need to manually edit your config. See the official documentation: https://docs.frigate.video/configuration/review/#restricting-alerts-to-specific-labels |
|
Did that change recently? Because i'm sure i didn't have it before 17.0 made it simple by adding another Zone and alow Alerts only in this zone. |
|
No, there were no changes to Review behavior in 0.17. |
|
Also, don't forget that presence in a zone is based on the bottom center of an object's bounding box. It is completely possible that the person in your screenshot above was detected differently than you expected. The Tracked Object Details pane in Explore or the Detail pane in History will give you more info. |
|
That was probably the issue... Increased zone size... Maybe someone touched the PTZ camera controls a little as that was no issue before ^^ |
|
Thank you so much for your great work! |
|
my understanding is that it is just regular AVX |
|
Thanks, I just did the update on my regular AVX server and so far it's running fine with no issues. I haven't tried the new model training feature though so I'm not sure about that one. |
|
Oh, I am running Frigate on a system with Intel Celeron J4105 together with Coral TPU, so no AVX means no Frigate 17? :( |
|
If the new state/object classification features are not used - will 0.17.0 run on a CPU without AVX? |
|
There are other features that require it too |
|
I may be alone here, but Frigate will not support nVidia Blackwell GPUs until 0.18. For those who have a Blackwell here's the solution for now. I run Debian 13.3 on an AMD Ryzen system in a KVM VM with GPU passthrough, and use Podman to build the containers (not the unsecure Docker). I split out the ORT (ONNX Runtime) container as it is a long build, and frees you to run many iterations of the Cormorant (née Frigate) build. The only thing you have to install in the VM is the nVidia driver v590+, podman, and podman-compose. For nVidia driver you must install using the .run method. I do both container builds as my unprived user. (not even allowed sudo {shudder} - pardon my infosec-awareness). I put my two build files in ~/src/cormorant/. Architecture Overview Final Target Stack (Latest Everything): Build podman.ort wheel factory so: ORT container is a wheel factory in the truest sense — its only job is to compile ORT from source with my exact CUDA 13.1 + TensorRT flags, and output a .whl artifact. The cormorant containerfile then reaches into it, grabs that wheel, and installs it permanently into the Cormorant image. A useful mental model: podman.cormorant At runtime you only ever run one container — cormorant. I can provide details on running and a systemd file if any interest. The ORT factory image just sits there in your local Podman image store, never running, only consulted again if you need to rebuild. You could even podman image rm it after a successful cormorant build and nothing would break. Build podman.cormorant so: Words to the wise:
Other shit you need: a Model (which holds the 'weights', ie 'wisdom' of your AI), a compose.yml and config.yml. The basic model is yolo9, but you can add tools, weapons, fire/smoke recognition, etc from Roboflow Universe. Comments, criticisms, insults, welcomed. |
|
I see. Blackwell users haven't reported any difficult issues for us so far, and the dev builds are stabilizing nicely. We'll see what comes up during the beta period. |
|
I know. It's alot easier when everything else is ancient and inappropriate for outdoor security work. Beta may exonerate you in that case. And pardon me hawkeye, you are not the one I am targeting. I am reluctant though as it may benefit NickM_"I would suggest you read the documentation..."-27 |
|
I'd like to offer a different perspective. Nick has helped me multiple times on Reddit and has always been patient, helpful, and generous with his time. My experience with him has been nothing but positive. Open-source maintainers often receive far more criticism than thanks despite volunteering countless hours to build and support projects like Frigate. So I'd just like to thank Nick and the rest of the maintainers for the work they do. Frigate is an excellent project, and the effort is appreciated. |
|
My experience is different. Vote me down as much as you want, Idgad. I started here with the best of attitude and intentions and got whacked, with zero justification. I will not have it and screw NickM-27. I always just take when presented with such circumstances. Like it or lump it. Do better than I have or eat the dog food. I am running the newest everything on 0.17.1. It is a good project but I am running some worthy refinements. We all make our choices. |
|
Please take your musings elsewhere. I'm not interested in reading your posts berating Nick or any of the other maintainers. I have read every post you have made from the beginning, and I don't think you are engaging within the bounds of our Code of Conduct. |
|
Update on the feedback: I've lowered the camera frame rate to 5, but the issue still occurs. Do you have any suggestions for debugging this?
|
|
I referenced the multithreading design of other modules, came up with a simple modification inspired by them (vibe), and the issue hasn't occurred since! This change is really effective! |
|
The idea in general seems fine, I think the main concern is that semantic triggers depend on the embeddings but I didn't take a close look at the code. |
I don’t think so, because I previously tried disabling semantic search, and the problem persisted. |
|
I think you misunderstand, I'm just saying we have to be careful about how this change affects feature order |
|
Update: After testing, there has indeed been significant improvement. However, if my CPU is running high-load tasks (such as training an object classification model), I still encounter cases where targets are lost in the results. My current modification only mitigates the issue but doesn’t completely resolve it. This problem is especially noticeable on low-power devices when multiple object classifications are configured. For now, I can continue using the modification I mentioned earlier, which at least greatly alleviates the issue when no model training is happening. I hope others can share their thoughts or suggestions. |
|
An update to my TrueNAS "app" to v0.17 went painlessly; now I've some new stuff to play with! Just in case, I took the precaution of stopping the app, backing up my three database files, and then upgrading. I didn't bother doing anything to my config (I am using git to control the versions I have); anyway it all works fine. Great work by the Frigate team as always. |
|
There is no switch in version 17? |
|
You will want to double check Frigate's connection to your MQTT broker as well as the Frigate integration version. |
|
it works, thanks |
|
I have been having issues since beta2 with facial recognition failing. I think I have finally isolated it to stopping motion/recording/detection while it is likely tracking a subject. I can consistently reproduce it. |
|
I can't reproduce this but we can catch this regardless so it doesn't crash. |
|
Good deal. Thank you. Were there changes in beta 3 that could be causing this or is it just a coincidence on my end? |
|
Perhaps there was something subtle, hard to say |
|
Can I limit the custom classification to a specific camera? That is, transfer the configuration from global to camera? |
|
Object classification can't be configured per camera but the updates are per camera already |
|
Awesome new features in this release; thanks! In the Classification screen, Is it possible to edit or choose which image is used for the object thumbnail? Can I rename a file in the clips directory or anything? Two of my custom objects have thumbnails that are very light/washed-out images, so it's a little difficult to read the white text label over it. A very minor issue; just curious. |
|
Thank you guys for your excellent work! |
|
is there a possibility to record the 247 stream in SD quality, and triggers/movements in HD, this is how i used to set up the v16 but now it doesn't work its or ?HD only or SD and HD all day long is 500Gb a day. also the cpu is busy, im running face detection on 2 cams HD, movement detection, and no more 247 36-41 Watt from wall , when using v16 247 record on SD, movement HD, face detection HD, and sym search the power was about 32/36Watt idle , so less options on more power use now but i assume of the new engine its running on ? |
|
Frigate has never supported separating SD and HD streams for recording. The recording configuration changed in 0.17 as the release notes indicated. You may want to review the recording docs and adjust your configuration accordingly: https://docs.frigate.video/configuration/record/ |
|
huh, that's strange i always thought i had it on sub stream SD recording for the 247, i mean when you use the slider to scroll you can see the sub stream record over the whole day right ? i thought you could set that on hd or sd. Thank for the fast reply software futher is great ! |
|
The low resolution scrubbing is done with previews, which are taken from the stream defined by the |
|
Very nice release 🚀 Thank you guys! |
|
I'd suggest creating a support discussion |
For some reason, even when the visual classifier is enabled, it still doesn't help much in certain scenarios.
|
I found an anti-interference algorithm called ByteTrack designed for multiple overlapping targets, which might be helpful for us. From what I’ve heard from industry insiders, some well-known security projects in China are already using this algorithm. |
|
Norfair is already built for handling this, the basic fact is with inaccuracies in object detection there will always be an error rate. The risk of moving to a new object tracker would have to have a major upside to deal with all the migration, potential bugs, etc. with all the features that already utilize tracking. We also are not using the Re-ID feature of norfair currently as models did not exist to be fast enough last we checked |
|
Frigate's Norfair integration is heavily customized for our specific use case - it includes per-object-type Kalman filter tuning (separate configs for cars, license plates, PTZ persons), a custom 4D distance function that normalizes position change by box size, histogram-based visual re-identification for PTZ autotracking, and optical-flow camera motion compensation via Norfair's |
What are the performance benchmarks? Must it be run on the CPU? I noticed that auto-tracking appears to use ReID, and I’m wondering how demanding this is on performance in multi-object scenarios. |
|
Autotracking is using a very simple histogram based Re-ID as a supplement that has proven to be helpful in some limited scenarios. It certainly isn't robust enough for anything else. I've done a lot of investigation into Re-ID models, and as @NickM-27 mentioned, none of them are fast enough or robust enough to even consider implementing in Frigate. |
|
Feedback from the Chinese community indicates that most users haven’t found how to customize the camera layout. Currently, you must first create a Camera Group before being able to customize the layout. This guidance may need to be improved—for example, displaying a custom layout button on the default page but disabling it, and using tooltips or other prompts to remind users to create a Camera Group first. |
|
Can someone help me to understand the default for state classification? Per the docs it seems neither motion or interval are enabled, so does it run at all? |
|
If you manually configured it then yes, but the wizard enables motion by default |
|
👌🏼 thanks. |
|
Hi ! with 0.17.1, when a role only has access to selected cameras, the review page (tab Motion) still shows cards for all cameras. |
|
Thanks, this will be fixed in the next maintenance release. PR is here: #23294 |
|
Thanks ! |
|
The problem is AV1 is only supported by a very few AXIS cameras at this
point, unless you mean re-encoding all video to store as AV1, which seems
unnecessarily burdensome on the CPU/GPU. Hell, most people don't even turn
on H.265 because they don't know what it is.
Much of the back end is/was in python last I checked, so any other language
refactor seems low yield. That said, if you got that kind of skill set,
party on. I'm still waiting to see your model and dataset... :)
…On Sun, Jun 14, 2026 at 2:36 PM Quantum ***@***.***> wrote:
Whelp, my model is loading and running clean on Blackwell with four
patches -- selected YOLOv26 classes along with my custom ones, through the
parse path which used to be impossible.
0: Ambulance
1: Bicycle
2: Boat
3: Bus
4: Car
5: Cat
6: Courier
7: Dog
8: Drone
9: Fire
10: Fire Extinguisher
11: Fire Truck
12: Heavy Truck
13: Helicopter
14: Ladder
15: Meteor
16: Motorcycle
17: My Car
18: Neighbor Vehicle
19: Parcel
20: Person
21: Police Car
22: Projector
23: Rat
24: Smoke
25: Snake
26: Surveillance Camera
27: Truck
28: Weapon
29: Wheelchair
30: Wild Bird
31: Wildlife
I realize that I am only talking to myself since I am muzzled, but I pat
myself on the back just the same.
To the devs who care (one bad apple spoils the barrel), I suggest AV1 and
RUST for all. Maybe I'll do those at some point.
—
Reply to this email directly, view it on GitHub
<#22137?email_source=notifications&email_token=AA2D23GY7YRZS3J5MQXINAT473WCJA5CNFSNUABIM5UWIORPF5TWS5BNNB2WEL2ENFZWG5LTONUW63SDN5WW2ZLOOQXTCNZSHE4TOMBUUZZGKYLTN5XKOY3PNVWWK3TUUVSXMZLOOSWGM33PORSXEX3DNRUWG2Y#discussioncomment-17299704>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AA2D23AEOTJ6MCC7646HGUL473WCJAVCNFSNUABHKJSXA33TNF2G64TZHMYTMNZWHE2DCOJUHNCGS43DOVZXG2LPNY5TSNJUGI3DAOFBOYBA>
.
Triage notifications, keep track of coding agent tasks and review pull
requests on the go with GitHub Mobile for iOS
<https://github.com/notifications/mobile/ios/AA2D23FL6TJ74BGRVAKREXT473WCJA5CNFSNUABIM5UWIORPF5TWS5BNNB2WEL2ENFZWG5LTONUW63SDN5WW2ZLOOQXTCNZSHE4TOMBUUZZGKYLTN5XKOY3PNVWWK3TUUVSXMZLOOSVGM33PORSXEX3JN5ZQ>
and Android
<https://github.com/notifications/mobile/android/AA2D23GXWJ5ALDZWV43UQOD473WCJA5CNFSNUABIM5UWIORPF5TWS5BNNB2WEL2ENFZWG5LTONUW63SDN5WW2ZLOOQXTCNZSHE4TOMBUUZZGKYLTN5XKOY3PNVWWK3TUUVSXMZLOOSXGM33PORSXEX3BNZSHE33JMQ>.
Download it today!
You are receiving this because you commented.Message ID:
***@***.***
com>
|
|
I mean reencoding to AV1, because it is right, not because it is facile. It appears that few here have heard of it and less have the hardware. But I see the value even if most don't understand the difference. I intend to archive for a long time. Honestly, converting to RUST is an unknown to me at this point. Maybe I will try, maybe I won't. But if you do not set goals... if you do not fail and keep trying... you are not DOING ANYTHING with your life and are worthless. Volunteer your fat ass to be put in the furnace to provide heat for the rest of us, giving at least some value. |
|
I disagree with re-encoding, for a number of reasons, but that's just me.
The incremental savings just isn't worth it to me unless you're recording
continuously on low/medium activity scenes, and I'm not doing that.
Besides, AV2 is coming out soon, so why not shoot for that. I'm betting
the FFMPEG crew will pick it up pretty quick.
…On Sun, Jun 14, 2026 at 6:19 PM Quantum ***@***.***> wrote:
I mean *reencoding* to AV1, because it is right, not because it is
facile. It appears that few here have heard of it and less have the
hardware. But I see the value even if most don't understand the difference.
I intend to archive for a long time.
Honestly, converting to RUST is an unknown to me at this point. Maybe I
will try, maybe I won't.
But if you do not set goals... if you do not fail and keep trying... you
are not DOING ANYTHING with your life and are worthless. Volunteer your fat
ass to be put in the furnace to provide heat for the rest of us, giving at
least some value.
—
Reply to this email directly, view it on GitHub
<#22137?email_source=notifications&email_token=AA2D23HFIIZUJJHNPO2WPPT474QI7A5CNFSNUABIM5UWIORPF5TWS5BNNB2WEL2ENFZWG5LTONUW63SDN5WW2ZLOOQXTCNZTGAYDSOBRUZZGKYLTN5XKOY3PNVWWK3TUUVSXMZLOOSWGM33PORSXEX3DNRUWG2Y#discussioncomment-17300981>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AA2D23DBUGFPENKVJMXCLJD474QI7AVCNFSNUABHKJSXA33TNF2G64TZHMYTMNZWHE2DCOJUHNCGS43DOVZXG2LPNY5TSNJUGI3DAOFBOYBA>
.
Triage notifications, keep track of coding agent tasks and review pull
requests on the go with GitHub Mobile for iOS
<https://github.com/notifications/mobile/ios/AA2D23F7ADRLIBGEQN3RB6T474QI7A5CNFSNUABIM5UWIORPF5TWS5BNNB2WEL2ENFZWG5LTONUW63SDN5WW2ZLOOQXTCNZTGAYDSOBRUZZGKYLTN5XKOY3PNVWWK3TUUVSXMZLOOSVGM33PORSXEX3JN5ZQ>
and Android
<https://github.com/notifications/mobile/android/AA2D23AQZZBB2G5LX5MKQNT474QI7A5CNFSNUABIM5UWIORPF5TWS5BNNB2WEL2ENFZWG5LTONUW63SDN5WW2ZLOOQXTCNZTGAYDSOBRUZZGKYLTN5XKOY3PNVWWK3TUUVSXMZLOOSXGM33PORSXEX3BNZSHE33JMQ>.
Download it today!
You are receiving this because you commented.Message ID:
***@***.***
com>
|
|
Fair enough. But to me it is not about compaction, it's about politics. And evidently AV2 dropped on 28 May and well, if there's a firmware update to the latest cards no objection. If not I'm Ok with AV1 for the time being. Ahh... Now I've offended all the fatties... Oh well. |
|
Nah, it won't be a firmware update, most likely. It's almost always a
hardware upgrade (at least for encoding). The older Ampere cards don't
even support AV1 encoding.
It'll be a while before we see it in anything I suspect.
…On Sun, Jun 14, 2026 at 6:47 PM Quantum ***@***.***> wrote:
Fair enough. But to me it is not about compaction, it's about politics.
And if AV2 is around the corner well, if it's a firmware update to the
latest cards no objection. If not I'm Ok with AV1 for the time being.
—
Reply to this email directly, view it on GitHub
<#22137?email_source=notifications&email_token=AA2D23DHT6N3WZCJHH5ANBT474TQBA5CNFSNUABIM5UWIORPF5TWS5BNNB2WEL2ENFZWG5LTONUW63SDN5WW2ZLOOQXTCNZTGAYTAOBZUZZGKYLTN5XKOY3PNVWWK3TUUVSXMZLOOSWGM33PORSXEX3DNRUWG2Y#discussioncomment-17301089>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AA2D23GC3ITICBOYIJZ7TSD474TQBAVCNFSNUABHKJSXA33TNF2G64TZHMYTMNZWHE2DCOJUHNCGS43DOVZXG2LPNY5TSNJUGI3DAOFBOYBA>
.
Triage notifications, keep track of coding agent tasks and review pull
requests on the go with GitHub Mobile for iOS
<https://github.com/notifications/mobile/ios/AA2D23EB4PSZMWBCV7Y2UJT474TQBA5CNFSNUABIM5UWIORPF5TWS5BNNB2WEL2ENFZWG5LTONUW63SDN5WW2ZLOOQXTCNZTGAYTAOBZUZZGKYLTN5XKOY3PNVWWK3TUUVSXMZLOOSVGM33PORSXEX3JN5ZQ>
and Android
<https://github.com/notifications/mobile/android/AA2D23BKTMJZYN632KZZJLD474TQBA5CNFSNUABIM5UWIORPF5TWS5BNNB2WEL2ENFZWG5LTONUW63SDN5WW2ZLOOQXTCNZTGAYTAOBZUZZGKYLTN5XKOY3PNVWWK3TUUVSXMZLOOSXGM33PORSXEX3BNZSHE33JMQ>.
Download it today!
You are receiving this because you commented.Message ID:
***@***.***
com>
|
|
You're right. A firmware or driver update won't bring AV2 to the 5050. NVDEC and NVENC are fixed-function silicon — the decode/encode logic is hardwired at tape-out, not microcode you can reflash. The consensus is software-first: through 2026 browsers and streaming platforms integrate the reference code while silicon makers are expected to add hardware decoders in 2027–2028, and some analyses push realistic consumer rollout later still, partly because AV2 is roughly five times more complex to decode than AV1. So it's not a concern for me at this time. I'll be dead by the time AV2 comes out. |
and others
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
Images
Changes since RC3
Major Changes for 0.17.0
Breaking Changes
There are several breaking changes in this release, Frigate will attempt to update the configuration automatically. In some cases manual changes may be required. It is always recommended to back up your current config and database before upgrading:
frigate.dbfilegenaiconfig now only configures the provider. Other fields have moved underobjects -> genai. See the new GenAI documentation.record -> continuousandrecord -> motionare separate config fields. See the examples in the documentation.smallmodel, which performs well on both CPU and GPU. Thelargemodel is the same as 0.16's and is not as accurate as the upgradedsmallmodel in 0.17. Uselargeonly if you live in a region with multi-line plates and you are having issues detecting text on them with thesmallmodel.detectresolutionwidthandheightfor cameras in your config if Frigate hangs on startup.exec,expr, andechosources for go2rtc are now removed by default to reduce the security risk if an attacker has access to the configuration. This can be disabled using an environment variableGO2RTC_ALLOW_ARBITRARY_EXECA separate configuration for this for HA addon users will come in a later beta. See the documentation.New Features
Frigate 0.17 introduces several major new features.
Classification Model Training
Frigate 0.17 supports classification models in two separate types: state classification and object classification. These models are trained locally on your machine using
ImageNetviaMobileNetV2.State Classification
State classification allows you to choose a certain region of camera(s) with multiple states, and train on images showing these states. For example, you could create a state classification model to determine if a gate is currently open or closed.
See the documentation.
Object Classification
Object classification allows you to choose an object type, like
dog, and classify specific dogs. For example, you can train the model to classify your dogFidoand add a sub label, while not labeling unknown dogs. Another example would be classifying if a person in a construction site is wearing a helmet or not.See the documentation.
Custom Viewer Roles
Frigate 0.17 now has the ability to create additional viewer user roles to limit access to specific cameras. Users with the
adminrole can create a uniquely named role from the UI (orauth --> rolesin the config) and assign at least one camera to it. Users assigned to the new role will have:viewerrole has access to (Live, Review/History, Explore, Exports), but only to the assigned camerasSee the documentation.
Review Item Summary with GenAI
Frigate 0.17 supports using GenAI to summarize review items. Unlike object descriptions which add a searchable description, review summaries have a structured output that instruct the AI provider to generate a title, description, and classify the activity as dangerous, suspicious, or normal.
This information is displayed in the UI automatically making it easier to see when activity requires further review and easier to understand what is happening during a particular video segment.
See the documentation.
Semantic Search Triggers
Triggers utilize Semantic Search to automate actions when a tracked object matches a specified image or description. Triggers can be configured so that Frigate executes a specific actions when a tracked object's image or description matches a predefined image or text, based on a similarity threshold. Triggers are managed per camera and can be configured via the Frigate UI in the Settings page under the Triggers tab.
See the documentation.
Object Detector Improvements
Frigate 0.17 brings performance increases for many detectors as well as support for new object detection hardware.
Nvidia GPU Performance
Support for Nvidia GPUs has been enhanced by implementing CUDA Graphs. CUDA Graphs work to reduce the involvement of the CPU for each inference, leading to faster inference times and lower CPU usage. CUDA graphs do have some limitations based on the complexity of the model, which means that YOLO-NAS, Semantic Search, and LPR models are not accelerated with CUDA Graphs. They will still continue to run on GPU as they did before.
Intel OpenVINO
Frigate 0.17 supports running models on Intel NPUs, for many models performance on NPU is similar to GPU but more efficient, leaving room to run more enrichment features on the GPU.
OpenVINO has also had many optimizations put in place to reduce memory and CPU utilization for object detection.
RKNN
Frigate 0.17 brings several improvements to RKNN platform including:
largemodel sizes.Apple Silicon
Frigate 0.17 supports running object detection on Apple Silicon NPU. This is provided through the Apple Silicon Detector which runs on the host and connects via IPC proxy to Frigate, providing fast and efficient inferences when run within the same Apple device.
See the documentation.
YOLOv9 on Google Coral
Frigate 0.17 supports running a quantized version of
YOLOv9on Coral devices, bringing improved accuracy over the defaultmobiledetmodel. Note that due to hardware limitations, only a subset of the objects on the standard COCO labelmap is included. Frigate+ has also added support for YOLOv9 models on the Google Coral and includes support for all 41 Frigate+ labels.See the documentation.
New Community Supported Detectors
Frigate 0.17 has community support for several new object detectors:
Frontend Improvements
In addition to supporting the new features, the frontend has many improvements.
Detail Stream
History view in 0.17 supports an additional view mode, Detail. This mode shows a card for each review item, and expanding a card reveals all tracked objects and their lifecycle events. Selecting any lifecycle event seeks the video to that exact timestamp. You can also overlay a tracked object's path on the video to help with debugging.
Redesigned Tracked Object Details pane
The Tracked Object Details pane in Explore has been redesigned to streamline the layout and consolidate related information. The Object Lifecycle tab is now the Tracking Details tab, which displays video overlays of the tracked object instead of static images, giving a clearer and more intuitive view of its activity.
Revamped Settings
Frigate 0.17 has a revamped Settings menu with a sidebar that categorizes the available options. This brings more scalability which will make it easier to support full UI configuration in a future version.
NOTE: The Debug view has been moved to the single camera Live view instead of Settings. Access the Debug view by enabling the switch under the Live view settings (cog icon) menu.
Add Camera Wizard
Frigate 0.17 supports adding camera via the UI without manually modifying your configuration file. When installing and starting Frigate for the first time, the main dashboard will include a button to start adding cameras via the Wizard.
Access the Wizard from the
Cameras --> Managementpage in Settings.Update Without Restarting
Frigate 0.17 supports saving many more features dynamically. Cameras, zones, and masks will not require a restart to take effect when saved through the UI. More will come in future versions.
Configuration Safe Mode
If an invalid configuration is detected, Frigate will enter safe mode and highlight the location of the issue. While in safe mode, the frontend is limited to the configuration editor, making it easy to correct the problem directly in the UI without needing an external file editor.
Other Notable Frontend Improvements
Other Backend Features and Improvements
Audio Transcription and Analysis
Frigate 0.17 supports fully local audio transcription using either
sherpa-onnxorfaster-whisper. The single camera Live view in the Frigate UI supports live transcription of audio for streams defined with the audio role, and anyspeechevents in Explore can be transcribed and/or translated through the Transcribe button in the Tracked Object Details pane.See the documentation.
Process and Efficiency Improvements
Frigate 0.17 uses the forkserver spawn method, this allows for better segmented memory control and better process management. Some processes are also started with lower priority, allowing the most important processes to have more CPU time when it is required.
Review Item Improvements
Review items have been refined to behave more intuitively:
Revamped stationary object tracking. Stationary object tracking has been enhanced to use new features to reduce incorrectly marking objects as active:
Smarter handling of loitering objects. Stationary behavior is now dynamic based on object type. Objects that are normally stationary for long periods (e.g., cars) will no longer keep a review item active indefinitely when stopped inside a loitering zone. Objects that are not expected to remain still (e.g., people) will continue the review item as long as they stay within the zone.
Severity-based review item cutoff. Review items now end when a higher-severity event (such as an
alertfor arriving home) finishes. Ongoing lower-severity motion (e.g., passing cars) will no longer keep the higher-severity review item alive. In these cases, thealertends and a newdetectionreview item begins immediately.Enrichment Improvements
-,deviceconfig option. This is useful in cases when multiple GPUs are available. DocumentationOther Improvements
networking -> ipv6 -> enabled. DocumentationThis discussion was created from the release 0.17.0 Release.
All reactions