Google is rolling out a redesigned document scanner in Google Drive and Files by Google on Android, introducing Smart Batch Scanning, Auto-Best Frame, and Duplicate Detection, per reporting from Android Authority, 9to5Google, Android Police, NokiaPowerUser, and BigGo. The feature set runs entirely on-device via Google Play services and requires Android devices with at least 8 GB of RAM, according to 9to5Google and BigGo. Sameer Samat, president of the Android ecosystem, shared a demo on X showing continuous hover-to-capture scanning, automatic splitting of pages into separate documents, and UI changes including a Material 3 Expressive viewfinder, per Android Police and Android Authority. The rollout is reported as happening now for eligible Android devices.
What happened
According to reporting by Android Authority, 9to5Google, Android Police, NokiaPowerUser, and BigGo, Google is rolling out a redesigned document scanner in Google Drive and Files by Google on Android. The update introduces Smart Batch Scanning, which lets users hover over multiple pages to capture them continuously and have the system automatically split pages into separate documents (Android Authority; 9to5Google). The release also includes Auto-Best Frame for swapping out blurry frames and Duplicate Detection to skip repeat captures (9to5Google; Android Authority).
Sameer Samat, president of the Android ecosystem, shared a screen recording and feature list on X that demonstrates the hover-to-capture workflow and previews appearing at the bottom of the viewfinder (Android Police). Multiple outlets report the UI now uses a Material 3 Expressive viewfinder and that the functionality is delivered via Google Play services with processing happening entirely on-device (9to5Google; BigGo).
High-stakes technical and availability details reported by 9to5Google and BigGo include a minimum device requirement of 8 GB of RAM and distribution via Google Play services, which also enables the scanner to appear in third-party host apps such as Files by Google (9to5Google; BigGo).
Technical details
Editorial analysis - technical context: The publicly reported feature set relies on on-device computer vision and frame-selection logic rather than server-side processing. Multiple outlets state the processing runs via Google Play services, which enables the scanner to appear in other host apps such as Files by Google (9to5Google; Android Authority; BigGo). Reported capabilities include continuous frame capture with a best-frame selector and duplicate-detection heuristics. These are typical mobile CV tasks implemented with lightweight models and heuristics to maintain latency and battery characteristics on-device.
Context and significance
For practitioners, this update illustrates a broader industry pattern of moving inference to the edge for latency, privacy, and offline availability. On-device scanning that performs frame selection and deduplication reduces reliance on network connectivity and cloud upload for initial quality processing, which aligns with trends in mobile ML where SDKs and platform services host optimized models.
What to watch
Observers tracking adoption should watch for:
- •broader device eligibility beyond the reported 8 GB RAM flo - •how Google surfaces the scanner to other apps via Play services
- •any developer-facing APIs or documented intents that would allow third-party apps to embed the same workflow
Monitoring user feedback and performance benchmarks will show whether on-device processing meets real-world latency and battery expectations reported by early reviewers.
Practical note for users
Reporting indicates the feature is rolling out now to eligible Android devices and is available where Google Play services and the updated Drive and Files apps are present (Android Authority; 9to5Google; BigGo). Sameer Samat's public demo on X provides a visual walkthrough of the new hover capture and preview UX (Android Police).
Scoring Rationale #
This is a notable product upgrade with practical utility for many users and mobile practitioners. It exemplifies the edge-processing trend and could influence how mobile CV features are delivered, but it is not a frontier-model or industry-shaking release.
Practice interview problems based on real data
1,500+ SQL & Python problems across 15 industry datasets — the exact type of data you work with.