Fixed an issue where browsing history results could update too early while typing in the address search field. Filtering now waits until you pause typing.

This update makes signing in smoother and improves accessibility. • Fixed an issue where the sign-in screen could stay open after signing in. • Improved VoiceOver navigation on the Welcome and subscription screens. • Fixed the Welcome screen appearing again for some users after updating. • Clarified how to set up Mail when sending a problem report.

• A clearer welcome experience and smoother guided tour, including improved highlights on Mac. • More reliable map findings, keeping saved images, page addresses, and GPS locations aligned. • Safer model replacement that preserves your existing model if an import fails. • Clearer messages when a model import needs more memory. • Confirmation before deleting models or clearing website data. • Fixed a crash during background message checks.

* Share a specific location (url and gps) of an object via a link or QR code * Log in to store your findings in the cloud * Organize and manage your findings * Detect web content using different tiling methods

A major update, and the first in a while. You can now save findings from any page, then export or share them. Classification runs alongside detection, and getting models is far easier: browse and install them from in-app catalogs, or import your own ONNX model and have it converted on device — labels, preprocessing and licence details are detected for you. The browser has been rebuilt around fullscreen use, with a Safari-like adaptive address bar, and going offline no longer clears the page you were reading — detection keeps working on it. New installs now get 14 days of free access to everything, with no account and no card. Signing in is entirely optional and only used to carry findings between devices. Also: an interactive onboarding tour, more languages, in-app problem reporting, and a great many fixes.

It is possible to download a mlmodel from a web address. Tap the CPU button in the top left corner, tap the plus button, select the globe, and enter the model's address. Once the model is downloaded, it can be imported in the application.

- The whole screen covers now the area where the objects are being detected, before v1.0.4 it was a rectangle in the centre. - Last page and default page handling fixed - Browsing history form updated

- better address bar and history links handling - fixed colors in the intro slideshow - fixed loading of the demo detect models - minor fixes in slideshow, texts