A deep dive into the world of mobile ML, including industry highlights, key use cases, opportunities and challenges, and more.
Download ReportAllow Snapchatters to engage directly with brand logos, product packaging, and other unique environmental elements.
Recognize, track, and isolate more components of a real-world scene.
Work with custom brand assets, objects, or other elements not found in common image datasets
Increase interactivity between the physical environment and digital objects and effects.
More effectively implement collision and boundary detection for custom objects.
No need for expertise with Python notebooks, model converters or other model-building overhead.
All the ML magic happens inside Fritz AI. Export high-performance models directly to Lens Studio.
On-device Dataset Collection
Synthetic Data Generation
Dataset Import
Performance Benchmarking
Model Retraining
Model Training
Real-world Data Collection
Model Delivery Network
Model Protection
Real-world Performance Data
Model Delivery Network
Device Analytics
Avoiding network round-trips and data shuffling to the cloud results in brilliant performance.
ML on-device performs consistently on a few devices or a few million, unlike cloud-based solutions.
Cloud-based providers of ML are expensive. Using local on-device processing saves money.
Personal and identifying information does not need to be sent to the cloud, ever.
All Fritz AI solutions work fully on-device and offline for maximum portability (even on airplane mode).
From photos to current whereabouts, useful data can stay entirely local and never touch the cloud.