The newest version of Google’s ML Kit includes three important new APIs (Application Programming Interfaces). The On-Device Translation API makes it possible to develop apps that can translate simple texts on a smartphone even without connecting to a cloud service. The only limitation: One of the two languages – target or source – always needs to be English. Although translations can also be done between other languages (Google offers 59 in total), English always needs to be used as an intermediary step. However, it is possible to scan the text using the camera and other APIs, then automatically determine its language and translate it.

The Object Detection and Tracking API has also been launched. This can identify the object in question in an image, track it in real time and then detect it in other sources. AutoML Vision Edge, another new API, ultimately assists in the recording of training data that can be used by applications that classify images.