In a previous article — Build iOS-ready machine learning models using Create ML—we learned how to create and train our own Core ML model using Xcode.
Then, we learned how to fine-tune these custom models to provide accurate results. For that, check out the following tutorial: Create ML — Increasing ML model accuracy.
Continue reading “Building a multi-class image classifier on iOS”
At WWDC 18, Apple announced Create ML, a platform to create and train custom machine learning models on your Mac. You can train models to perform tasks like image recognition, extracting meaning from text, or finding relationships between numerical values.
Continue reading “Working with Create ML’s MLDataTable to Pre-Process Non-Image Data”
One of the coolest things for me about our team’s work with Lens Studio is that I get to learn while being creative—and being silly as well.
In all honesty, Snapchat has long been one of my favorite apps, if for nothing other than a lone filter that weirdly changed the course of my life…
Continue reading “Snapchat Lens Recipe: Which ____ Are You?”
Capturing document images with the help of handheld devices is the norm today. These images can easily be transferred over mail, chat, and can also be stored in the hard drive for future reference.
However, casual capturing of documents often leaves behind folds and distortions that are hard to recover from using normal editing processes.
When I first saw CyreneQ’s lenses, I was immediately struck by how fun and quirky each individual one was. From bearded dragons hanging out on your face, to a literal croc on your foot, each one is joyful, unique, and ultimately shareable.
I caught up with Cyrene, who’s used Snapchat since its inception, to learn more about her inspiration. Some highlights of our conversation are below.
Frank and Xuan of Paper Triangles really enjoy and are enthusiastic about the work they do—that was one of my main takeaways from my recent conversation with them.
Their Lenses reflect that spirit—they’re often vibrant and atmospheric, with a generally light color palette that sometimes verges on something resembling neon.
I also appreciate how they use old ceramic statue heads to display their Lenses on their Creator profile—adds a sense of elegance and grandiosity that complement their aesthetic nicely.
Continue reading “Snapchat Lens Creator Spotlight: Paper Triangles”
Cross-platform mobile app development had always been a dream for both iOS and Android app developers. Even with the existence of tools like Cordova and Ionic, this dream has long remained elusive, as these tools resulted in performance that was no where near native performance.
Many argue that React Native is the best framework for the same, but ever since the launch of Flutter, this has started to change.
Continue reading “Building a Cross-Platform Image Classifier with Flutter and TensorFlow Lite”
Following what we learned about human vision, we’re now going to transfer some of that knowledge to computer vision (exciting right?!).
Very quickly, to quickly explain/recap this series, Joseph Redmon released a set of 20 lectures on computer vision in September 2018. As he’s an expert in the field, I wrote a lot of notes while going through his lectures. I am tidying my notes for my own future reference but am posting them on Medium also, to hopefully be useful for others.
Continue reading “The Ancient Secrets of Computer Vision 3 by Joseph Redmon — Condensed”
Using ML Kit, the language of a string of text can be determined. The ML Kit API supports over 100 languages. It can also identify native and romanized text from languages such as Russian and Arabic. Given a string of text, the API provides the most likely languages as well as the confidence level. Let’s look at how that can be done.
Continue reading “Identify the language of text with ML Kit on Android”
Phased Long Short Term Memory is an improvement on the well known Long Short Term Memory units. Its main advantage is its ability to deal with data that do not follow a simple sequence and data with long timesteps.
Phased LSTM differs from LSTM by the possession of an additional gate called the time gate. The phased version is very efficient and performs better than basic LSTMs, even when given fewer data.
