Lobe started with the mission to empower anyone to build machine learning models and use them to solve complex problems that were once unimaginable with traditional programming. However, there are many steps involved in this process, and building your model is just the first step.
Today, Lobe 0.9 introduces Use to make it easier to find everything you need to put your model to work—with starter projects, improved export options, beginner tutorials, and brand new snippets and instructions for Lobe Connect, our local API. Starting and running your machine learning projects is now easier then ever.
Finding the best way to export and integrate your model can be difficult at first. And in previous versions of Lobe, export options were buried within a menu, making it seem like an afterthought that only needed to happen once. But exporting a model should be something you do repeatedly, so you can test and improve it over time. This new version of Lobe combines Play and Export into one tab called Use, the place you will go to test, export, and integrate your model.
Use showcases our improved local API Lobe Connect, all our export options, and starter projects from our GitHub, with links to more documentation so you can get expert guidance on how to put your model to work. We want Use to become the central place for all the present and future export options and integrations we will have. Let us know what options and integrations you’d like to see in future versions of Use.
We’ve been inspired by the many uses our local API has had, from analyzing creatures in a microscope, to setting up the room mood lighting and music based on vinyl covers. In this release, we wanted to make our API better by improving our code snippets, adding better instructions on how to use it with different languages, and making our endpoint more robust.
Lobe Connect lets you play with your model and improve it directly within Lobe by connecting to it without the need of exporting and integrating it to your app. If you want to learn more about it, please read our Exporting 101 series.
It can be hard to get a project started, but it’s incredibly rewarding when you get to see your model predict something in the real world. Because of that, we are introducing four new starter projects. An iOS app, an Android app, a REST server, and a Web app written in React. To get started, just export your model and drop it in the correct project folder.
We also wanted to continue to focus on fundamentals. This release introduces performance gains when switching between tabs, when loading and caching images, and when scrolling through Lobe in general. On top of that, we have also improved the performance of loading and running inference on exported ONNX models.
From the beginning, Lobe’s mission has been to empower anybody to solve their ideas with machine learning. We believe that this new version of Lobe gets us a step closer to that by introducing easy-to-use starter projects, an improved and rebranded API called Lobe Connect, and a Use tab to provide all the resources you’ll need to put your model to work.
We can’t wait to see what you do with this new version of Lobe. Download it for free to get started on your machine learning model today. And join the community to share your feedback and see what others are building.