On-device processing (AI)

On-device AI is repeatedly praised for speed and privacy, identifying people/vehicles/animals/packages (and sometimes faces) without cloud processing; occasional misclassifications still happen.

Several reviews highlight that detection runs on-device for faster, more private object recognition; Familiar Faces and some advanced intelligence features still require a subscription.

On-device AI enables fast detection without cloud latency and supports classification (people, vehicles, pets) and face features in some reviews. Accuracy is generally good but not perfect; a few reviewers saw false positives (shadows or small or large animals) and noted it improves with correction and updates.

People/pet/vehicle detection is handled on-device and is often described as free and effective, avoiding the common paywall for smart alerts.

On-device AI (with HomeBase) covers people/pets/vehicles and face recognition; it’s helpful for reducing review time, though some say face recognition is less accurate than Google’s best-in-class approach.

On-device subject recognition for people, pets, and vehicles is widely praised and helps reduce generic motion noise. Accuracy is generally strong, though performance still depends on placement and sensitivity settings.

On-device AI for person, pet, and vehicle detection is a standout and generally does not require a subscription; tracking can be strong, but a few report occasional tracking oddities.

On-device smart detection for people/vehicles/animals is a consistent theme, and alerts are often described as correctly classified. Tracking accuracy can vary with lighting and subject speed, but the core on-device filtering is generally well-liked.

On-device AI for human/vehicle detection and tracking is a core strength; advanced features like face recognition are typically positioned as HomeBase upgrades.

AI detection and auto-tracking are frequently highlighted, often described as impressive when tuned. Some reviewers report periods of over-detection that improve with updates and feedback.

AI features (face recognition plus person/vehicle/pet detection) are praised for reducing noise and improving alerts over time, especially in daylight. Multiple reviewers note accuracy can drop at night or in complex scenes and may require training/tuning.