Heartbeat Newsletter Vol. 46

Auto selfies on Pixel 3, what ML needs from hardware, a mobile ML resource list, image classification on Jetson nano, and more


Google introduces automatic selfies with Photobooth on Pixel 3

Joining “Top Shot” and “Portrait Mode”, the team over at Google has announced “Photobooth”, an AI-powered mode on the Pixel 3 camera app that detects 5 different expressions using two different on-device neural networks. The mode features an intuitive shutter mode that runs a series of quality assessments and analyses based on the predictions of the aforementioned models, all in real time. Lots more great discussion of how the tech works in the article. [Read More]

Smart home, machine learning, and discovery

In his latest article, Benedict Evans offers some insightful commentary and discussion about the historical context in which the notion of the “smart home” exists. He also digs into some of the primary questions surrounding how machine learning will interact with our lives, including: “how do we combine these commodity components into products that make sense — both in the home and on platforms and smartphones?” [Read More]

What machine learning needs from hardware

Jetpac CTO and technical lead of TensorFlow’s Mobile team Pete Warden explores what machine learning engineers need from their hardware from a software engineer’s perspective. He contends that there are a plethora of applications that are currently blocked from launching because there simply isn’t enough computing power. As such, he argues that the biggest need in the coming years will be hardware for inference, not training. [Read More]

[TensorFlow] MLIR: A new intermediate representation and compiler framework

One of the biggest challenges in machine learning is how difficult it is to optimize models for multiple platforms. Models are being deployed from the tiniest micro controllers to the largest TPU clusters. Getting things running quickly and efficiently on every chipset and platform is a tough job. To help, the TensorFlow team has released a new intermediate representation and compiler to take a single TensorFlow graph and generate compiled code optimized for any hardware. The project is just in its infancy, but we think it will be incredibly important down the road. [Read More]


fritzlabs / Awesome-Mobile-Machine-Learning

A curated list of resources for mobile machine learning, including: materials to get started, mobile ml frameworks, code and libraries, tutorials, and more. [Explore]

goodrahstar / my-awesome-AI-bookmarks

Curated list of reads, implementations, and core concepts of Artificial Intelligence, Deep Learning, and Machine Learning. A great resource for those looking to read up on both technical and non-technical AI/ML perspectives. [Explore]

mwleeds / android-malware-analysis

This project seeks to apply machine learning algorithms to Android malware classification. [Explore]

[Plugin Update] Flutter / tflite 1.0.3

A Flutter plugin for accessing TensorFlow Lite API. Supports image classification, object detection (SSD and YOLO), Pix2Pix, and Deeplab on both iOS and Android. [Explore]


How to run a Keras model on Jetson Nano

A step-by-step guide for running an image classification Keras model on the new jetson Nano. [Read More]

[Free Udacity Course] Core ML: Machine Learning for iOS

In this short course, you’ll learn how to incorporate Apple’s Core ML framework into your app. You’ll also get a quick overview of machine learning fundamentals, and exposure to real-world examples of companies using machine learning technology in their iOS apps. [Read More]

Building a Vision-Controlled Car Using Raspberry Pi — From Scratch

Learn how to build a car robot from scratch and control it using Raspberry Pi based on images captured by a USB camera. [Read More]

Writing a Simple Waypoint System in Augmented Reality on iOS

Learn how to write a simple waypoint system in AR on iOS that virtual objects will follow as they move through a scene. Think AR PacMan…[Read More]

Implementing ML Kit’s Smart Reply API in an Android App

ML Kit recently released two new APIs to perform on-device natural language processing. Here’s a quick implementation of the Smart Reply API on Android. [Read More]

read original article at https://heartbeat.fritz.ai/heartbeat-newsletter-vol-46-efca9fbb5dd5?source=rss——artificial_intelligence-5