Frigate NVR Updates: UI Improvements and Hardware Documentation
Four pull requests were merged yesterday focusing on user interface consistency, developer documentation, and hardware performance data. The changes include sorted class names in the web interface and updated RTX 5060 Ti inference benchmarks.
Duration: PT1M18S
Transcript
Good morning, this is your Frigate NVR update for March 14th, 2026.
Yesterday saw four merged pull requests with improvements across the platform. Hawkeye217 merged changes to consistently sort class names in the web interface, modifying the classification selection dialog and model training view to ensure predictable ordering. This affects 150 lines across two TypeScript components.
Leccelecce updated the developer contribution documentation, adding Python check commands that mirror the GitHub CI pipeline. Contributors can now run formatting, linting, and type-checking locally before submitting pull requests.
NickM-27 contributed two separate improvements. First, an update to the stationary object classifier that adds proper handling for license plates, which aren't expected to remain stationary like other tracked objects. Second, updated hardware documentation with RTX 5060 Ti inference performance data, including the new RF-DETR model and Nano-320 benchmarks.
An additional commit provided more detailed RTX 5060 Ti performance metrics, giving users better hardware selection guidance.
What's next: The focus on UI consistency suggests more interface standardization work may be coming. The expanded hardware documentation indicates continued support for newer GPU architectures.
That's your Frigate development update. I'm back tomorrow with the latest changes.