Once per quarter we meet with several members from the TensorFlow.js community to learn more about what amazing projects they have created. This thread is the one stop place to see the latest interviews to get inspiration of what is possible using Machine Learning in JavaScript across front end (browser), back end (Node.js), React Native (native app), Electron (desktop), and even IOT (Raspberry Pi via Node). Check back regularly!
To kick things off, here is our first video interview:
Enjoying the show with Gant Laborde who explains how he solved a problem when presenting digitally to an audience where he was unable to know if they were interested in the content being presented. Learn how Gant created an innovative, real-time, and scalable system to better understand his audience using machine learning in the browser using TensorFlow.js.
Real-time semantic segmentation in the browser with Hugo Zanini - a Python developer who was looking to use the latest cutting edge research from the TensorFlow community in the browser using JavaScript. Join us as Hugo takes us through his learning experiences in using SavedModels in an efficient way in JavaScript directly enabling you to get the reach and scale of the web for your new research.
In our next episode of Made with TensorFlow.js we head to Denmark to join Anders Jessen, who has been investigating powerful touchless interfaces powered by our TensorFlow.js hand pose model. Finally, our sci-fi-like interaction dreams can become reality!
In our next episode we head to Spain to join Cristina Maillo, to talk about her experience using machine learning for Yoga instruction. Cristina created an easy to use website using TensorFlow’s PoseNet that can guide you through your Yoga poses and time you as you hold each one! If you lose the pose the countdown stops and waits for you to readjust yourself.
Next we are heading to Amsterdam to join Charlie Gerard, a Senior Front End Developer at Netlify to talk about her latest creations. Join us as Charlie walks us through WashOS - a web based system that can detect how long you have been washing your hands for, and “splat”, a fruit ninja styled game powered by TensorFlow.js that enables you to use your hands and arms to chop fruit from anywhere you wish!
In our 6th episode of Made with TensorFlow.js we head to Australia to join Benson Ruan, who has used Natural Language Processing to understand the sentiment of tweets and is able to visualize the results. Now we can monitor in real time user sentiment as people react to any given topic.
In our 7th episode of Made with TensorFlow.js we head to India to meet Shivay Lamba who has created a virtual physio assistant to help you perform your daily exercises. With this system you can check you are doing the correct stretch using our PoseNet model live in the browser.
In our 8th episode of Made with TensorFlow.js we head to the USA to meet with James Seo who has created a visually stunning mixed reality demo to help us understand human pose over space and time that can be inspected from any angle you desire using WebXR.
On episode 9 of Made with TensorFlow.js we’re joined by Shan Huang from China, who’s built upon her previous Pose Animator project to make Scroobly, a fun app which brings doodles (SVG images) to life in real-time using your camera. Scroobly uses Facemesh and PoseNet to map your live motion and updates the animation as you move!
Next on Made with TensorFlow.js for episode 10 we’re joined by Chris Greening from the UK, who’s built an augmented reality web app to solve Sudoku puzzles in real-time. Chris breaks down his problem-solving techniques in building a complex app like this, including methods for image processing and character recognition.
This time on Made With TensorFlow.js we’re joined by Samarth Gulati and Praveen Sinha from India, to hear how they’ve used TensorFlow.js and Facemesh model to create a system that can recreate digital face masks based on cultural events around their country.
Today on Made with TensorFlow.js we’re joined by Andreas Schallwig from Shanghai. Andreas has been hacking on some pretty impressive demos for touchless interfaces on public smart displays such as photo booths and games. Check them out!
Today on Made with TensorFlow.js we’re joined by Emily Xie from New York, who’s managed to bring paintings to life using a combination of TensorFlow.js and TensorFlow Core.
Today on Made with TensorFlow.js we’re joined by Paul Jessop from London, who’s made some custom hardware powered by machine learning, that’s capable of tracking custom objects for sport videography and more - our very first madewithtfjs powered Kickstarter project!
In this episode of Made with TensorFlow.js we’re joined by Yining Shi and Bomani Oseni McClendon who are working on the ml5.js library that is built upon TensorFlow.js to try and make machine learning even more usable by everyone. From creative coding to hardware experiments, ml5.js can enable you to do many advanced things with just a few lines of code. Learn more and have a go yourself!
In this episode of Made With TensorFlow.js we’re joined by Kenny Song, an active researcher on security and reliability, where he shows you how to break neural networks in your web browser in real-time by changing inputs, such as pixels in an image, to fool a machine learning model. Watch as he turns a photo of a “dog” which is initially classified correctly to be misclassified as a “hotdog” - even though to you, as a human, the image still looks the same. Learn how he does it in this educational video so you can make your models even more robust to such attacks in the future.
This time on Made With TensorFlow.js we’re joined by Paul Van Eck, a software developer with IBM who shows how to use Node-RED, an open source visual programming tool that even supports machine learning with TensorFlow.js and can even deploy to a Raspberry Pi and more. Watch as Paul uses this system to keep his cat off the table, open his garage door when the correct car is recognized by its number plate, and more! Take command of the physical world with TensorFlow.js and Node-Red in this episode! Happy hacking.
Today on Made With TensorFlow.js we’re joined by Michelle Sun, an interaction designer, who solved a problem she had - never having a guitar tuner nearby when she needed one. Learn how Michelle created a system to tune any instrument (even your voice) live in the web browser using a pitch detection model known as Crepe without the need for any specialist hardware:
In this episode of madewithtfjs, Jason is joined by Director of AI at Dataiku, Vivien Tran-Thien who has used TenorFlow.js to create an impressive motion parallax effect with face tracking in browser. This allows you to view any 3D scene on your regular 2D screen as if you had a 3D monitor - no special glasses needed. By moving your head you automatically change the 3D scene’s perspective as it tracks your eye’s position giving the illusion of 3D to the viewer.
Join Rishit Dagli in this episode of Made With TensorFlow.js as he turns nighttime into daytime. Learn how he managed to convert cutting edge research from Python, specifically the MIRNet machine learning model, to run in the web browser via Node.js. Now anyone can see in the dark.