Google launched ML Kit, a new software development kit (SDK) for software app developers at its ongoing Google I/O 2018. This new SDK will make it easier for app developers to add artificial intelligence-powered features in their mobile apps. Google has opted to support different platforms with SDK as ML Kit is cross-platform. It means that users can add AI-powered smart to their apps regardless of the OS they are developing apps for which means that both iOS and Android app developers can take advantage of this SDK.
The SDK will include machine learning models for text recognition, barcode scanning, landmark recognition, image labelling and face detection that Google has already built, in their apps. What is cool is the fact that all these machine learning models are available both online, as well as online depending on what the app developer prefers.
Watch: How to install Android P Beta on select smartphones
Google plans to add machine learning models for smart replies and high-density face contour feature for face detection in the base set in coming months. Free to use offline machine learning models are likely to be a game changer as developers no longer need to worry about crafting AI-powered models from scratch or paying to use the service as reported by TechCrunch. However, the offline and online models will vary somewhat as offline models will be smaller in size and therefore less accurate while the online models don’t have such limitations in terms of size, increasing the accuracy dramatically.
For instance, the on-device models for image labelling are limited to 400 while the models on cloud extend beyond 10,000. Google added that ML Kit supports TensorFlow Lite models also for developers who want to go beyond the pre-trained models by Google. The ML Kit SDK comes under Google’s new Firebase mobile and web application development platform.