Now in Android #97. Gemini, AICore, ML Package, Android Studio… | by Alice Yuan | Android Builders | Dec, 2023

On December sixth , we introduced Gemini, our most succesful AI mannequin but!
It’s been optimized for 3 completely different sizes: Extremely, Professional and Nano. Gemini Nano is our most effective mannequin constructed for on-device duties, operating instantly on cellular silicon which opens assist for offline, privacy-sensitive inference.

We introduced that you’ll quickly be capable of use the Google AI Edge SDK to leverage Gemini Nano by means of Android AICore. It’s at the moment rolling out to Pixel 8 Professional with extra units and silicon companions to be introduced within the coming months.

AICore is at the moment getting used to summarize content material within the Recorder app, even when the cellphone’s community connection is offline.

In case you are desirous about constructing apps utilizing Gemini Nano and Google AI Edge SDK, please join our Early Entry Program.

The brand new ML Package Topic Segmentation API, now in beta, enables you to simply separate a number of topics from the background in an image. The API takes an enter picture and generates an output masks or bitmap for the foreground. It additionally gives a masks and bitmap for every one of many topics detected.

It allows cool use circumstances similar to background swap, or add distinctive results to folks, pets, or objects with only a few strains of code.

To get began go to the ML Package documentation.

Android Studio Hedgehog is now steady, and it’s packed filled with upgrades, similar to enhancements to Android vitals in App High quality Insights and the brand new Energy Profiler.

The Android SDK Improve Assistant now makes it simpler to improve functions to focus on Android 14.

Lastly, there are a variety of instruments to make it quicker to develop utilizing Jetpack Compose together with Preview’s Gallery Mode, Compose State data within the Debugger, and Compose Multipreview templates.

Now we have a number of new weblog posts so that you can try!

Learn extra on Digital Machine as a core Android Primitive — the Android Virtualization Framework (AVF) that can be obtainable on upcoming choose Android 14 units, and the way it will provide platform builders and privileged app builders new capabilities.

Additionally, learn extra on the launch of KSP2 Preview, which introduces assist for the brand new Kotlin compiler (code-named K2) and a brand new standalone supply generator that gives extra flexibility and options than the present Kotlin compiler plugin.

Lastly, study extra about how NordVPN boosted the pace of its login person stream by 60% utilizing Baseline Profiles.

Take a look at the video on Enhancing Jetpack Compose efficiency, the place Tomáš demonstrates strategies for efficiency by means of measurement, debugging, and strategic optimization.

The best optimization approach is conserving Jetpack Compose up to date to profit from the efficiency enhancements we implement underneath the hood. One other one is to allow R8 and generate a baseline profile, which may considerably optimize code execution. And thirdly, in the event you change state quickly in some circumstances, discover the choice to defer the compose section to stop pointless recomposition. Examine the video for a extra in-depth clarification.