Rerun

Visualize computer vision

5.0•2 reviews•

174 followers

Visualize computer vision

5.0•2 reviews•

174 followers

Visit website

Rerun is an SDK for logging computer vision and robotics data paired with a visualizer for exploring that data over time. It lets you debug and understand the internal state and data of your systems with minimal code. It’s built from the ground up in Rust to run fast everywhere.

Launch tags:Design Tools•Open Source•Developer Tools

Launch Team

Render: Get up to $100K in startup creditsThe fastest way to deploy and scale apps and agents

Promoted

Rerun

Maker

📌

Hello Product Hunt 👋 I’m Niko, co-founder and CEO of Rerun. We’re building the visualization stack for computer vision that I have always dreamt of as a developer. I met my two co-founders, Emil (the guy behind Rust GUI framework egui) and Moritz, when we were working at a 3D scanning for retail company called Volumental. Whilst working there I stumbled upon a talk by Bret Victor called Seeing Spaces. In it he made the argument that for a growing number of projects with embedded intelligence (robotics, drones, etc), the main challenge isn’t putting them together, but understanding what they are doing and why. I felt struck by this, someone was expressing what I was struggling with in my daily work as a computer vision developer. The next morning I showed up early to work and pitched Emil the idea to build a light version of a seeing tool. He got on it and two days later we had a prototype. Over time the tool became critical to the company’s success and was one of the main reasons why we brought a working product to market years before any of the competitors. After I left that company I have been missing this tool at every single project I’ve worked on. Whenever I found myself needing it I struggled with two problems that kept me from rebuilding something ambitious: 1. It’s a massive engineering effort to build a tool that is flexible, performant, allows you to explore over time, and can handle complex data like point clouds. 2. It’s actually not the main skill set of algorithm developers to build all the infrastructure that is required for a tool like this. These problems hold true across all of computer vision today. Even in the big tech companies you will find the most scrappy tools for visualization that leave an incredible amount of insight and productivity on the table. Therefore we decided to tackle this once and for all and started Rerun. What is Rerun? Rerun is an SDK for logging rich data together with a visualization tool for visualizing it. Over time, Rerun will evolve from a tool to a fully customizable toolkit, where you’ll have control over everything from layout to data transforms and shaders. You’ll even be able to embed single views inside other applications. What’s the tech behind Rerun? Rerun is built from the ground up in Rust to run blazing fast everywhere. The same code runs on your desktop, in the cloud, and in the browser using WASM. The renderer is custom built using high performance and portable wgpu. The in-memory data store is built on top of Arrow for flexibility and ecosystem support. The Viewer is built using an end-to-end immediate mode architecture using egui. How is Rerun licensed? The Rerun project is completely open source and dual licensed under MIT and Apache 2. We very much welcome contributors and you can join our nascent community on Discord and Github. Rerun the company is a for-profit enterprise. In the future, we will sell a product that builds on top of the open source project, and focuses on the needs of teams. I'm incredibly excited (and quite nervous) to finally share the Rerun beta with the world. I want to give a shoutout to all of our early testers whose feedback already has improved Rerun a ton. And to my incredible team that has been shipping non stop to make this launch happen.

Report

3yr ago

Sketchfab

Hunter

📌

Hi Product Hunt o/ Happy to be sharing Rerun today. I met the Rerun crew back in their Volumental days, not many teams really shipped computer vision products that worked out in the real world like they did! They are now applying all this experience at Rerun, working on a problem that everyone working in computer vision can relate to.

Report

3yr ago

BloggerAI

Congratulations on your launch on Product Hunt! 🎉 This sounds like a really powerful tool for computer vision developers and I'm curious to know how it's been used to tackle challenging projects. Could you tell us about some of the use cases you've seen? 🤔 I'm sure this will be a great help for many developers out there. Wishing you all the best and hoping to see more innovations from you soon! 🤩

Report

3yr ago

Rerun

Maker

@paolo_degiglio Thanks, @paolo_degiglio! It is still early days, but our alpha users included a really cool tennis analytics company (https://www.rerun.io/blog/playre...) and a company building navigation tools for the blind (https://www.rerun.io/blog/biped)

Report

3yr ago

BloggerAI

@moritz_schiebold That's look amazing 🚀

Report

3yr ago

Rerun

Maker

@paolo_degiglio Thanks 🤩

Report

3yr ago

Sequoia

Hey there! Congrats on the launch of Investor Rerun on the Product Hunt! 🎉 Wow! It looks fantastic! Do I have to know any programming language to use Rerun? If so, which one?

Report

3yr ago

Rerun

Maker

Thank you @denis_galka! You don't have to know how to code to use the Rerun Viewer but someone needs to know how to code to use the SDK to pass data to look at. The api is very simple though and can be used from Python so hopefully it's still relatively easy to get started.

Report

3yr ago

EnVsion AI

Congrats on launching Rerun Nikolaus! I don't do much computer vision these days but this is the type of tool we would have DIED to have when I was into mobility and self-driving cars a few years earlier. ReRun looks super sleek and smooth performance wise. Really impressive work 💪🏿 Are you also planning to help with data labelling in the future? I remember it was a huge pain at the time too.

Report

3yr ago

Rerun

Maker

Thanks @ed_forson ! We're not building Rerun as a data labeling tool but we know labelling is an important part of the workflow when building computer vision and ML products. In the future you should definitely be able to annotate and leave comment on your data inside Rerun. That would likely be implemented through the plugin system we're building. However we don't have the ambition to build a real labelling tool. A good labelling tool needs to help you with project management and structuring larger labelling projects and we think that is best left to dedicated tools that we will happily integrate with instead.

Report

3yr ago

EnVsion AI

@nikolaus_west Got it. It's great that you guys are staying focused on what you believe Rerun is uniquely position to do best. I'll certainly check it out when I do some computer vision stuff again.

Report

3yr ago

Carbone

Wowowow your video tool is incredible! Why does your project not have more votes?? Can we also mention it is open-source? Anyway, I'll have to give it a try!! Bravo @nikolaus_west !! You got a star and new follower on Github

Report

3yr ago

Rerun

Maker

Thank you @steevep! I really appreciate that! Who knows about the votes but I’m happy for the Github love 😊

Report

3yr ago