Blog

Matroid’s computer vision software is now being used to analyze home insurance claims. Everything from stress cracks in a home foundation, to water damage, damaged countertops, and more can be detected and categorized using Matroid.

Use Matroid’s computer vision software in conjunction with existing cameras to identify safety risks in the workplace; such as employees standing too close to running machinery, improper performance of procedures, or employees not wearing the proper PPE. Matroid can recognize other safety hazards such as blocked exits, slippery floors, precariously stacked objects, and more.

Use Matroid’s computer vision software in conjunction with existing systems to pick up defects in production processes. With Matroid, companies can see stages of assembly and defects more efficiently and reliably than the human eye. Combined with Matroid’s on-cloud deployment feature, errors are instantly flagged, followed by an automated and coordinated response to a PLC.

Check out this exclusive interview with Chris Hadfield led by Reza Zadeh, Founder and CEO of Matroid on Space and AI!

Nobody likes graffiti scattered across whichever beautiful city they live in. Whether it be on a box truck, train, wall, or street sign, it is visually unappealing, and it has become a symbol of impoverished areas. Matroid’s enterprise computer vision solution can detect graffiti wherever it may be. Check out how Matroid built a detector for graffiti here!

Matroid is proud to present ScaledML Talks from subject matter experts and world leaders. Learn from the best minds in the machine learning community.

Matroid is thrilled to be named a top company in Edge & IoT AI and CV. 

Matroid continues to be recognized internationally as the major player in Computer Vision. Come read more!

Matroid is proud to announce we have been awarded contracts totalling more than $20 million by the Department of Homeland Security and the Air Force. We are working hard to deliver high-quality software, in an ethical and socially responsible manner.

Matroid can build detectors across the visible spectrum. One of the best use cases for Matroid's Enterprise Computer Vision Solution is detecting suspicious objects on X-Rays. Check out how Matroid built a detector for suspicious objects in X-Ray imagery!

Computer Vision is the core of Matroids offering, but what exactly is Computer Vision? Read here to find out!

Our employees happiness is one of our utmost priorities here at Matroid. With our rapidly growing team we’ve moved into a new headquarters location that matches our ambitions in computer vision.

Petrographic analysis is a tedious, "rocky" and necessary exercise of geological analysis - using Matroid, our users can build petrographic detectors in minutes that allows them to automate inspection of samples.

Cracks are one of the earliest indicators of degradation of a structure or product. Using Matroid, our users can build a crack detector in minutes that allows them to automate detection of cracks and say no to cracks. 

CB Insights today named Matroid as the sole computer vision company to the fifth annual AI 100 ranking, which showcases the top 100 most promising private artificial intelligence companies in the world.

Computer vision makes routine wind turbine inspections safer and more efficient.

We are proud to announce that Matroid has raised $20M in Series B funding led by Energize Ventures with participation from existing investors New Enterprise Associates (NEA), Intel Capital, and others. The new financing brings our total funding to $33.5 million.

The new partnership between Matroid and Eagle Eye Networks delivers more advanced AI to Eagle Eye Cloud VMS customers.

The creators of TensorFlow, Kubernetes, Apache Spark, Keras, Horovod, Allen AI, Apache Arrow, MLPerf, OpenAI, Matroid, and others will lead discussions about running and scaling machine learning algorithms on a variety of computing platforms, such as GPUs, CPUs, FPGAs, TPUs, & the nascent AI chip industry.

In a collaboration with Stanford Hospital and hospitals in Hong Kong, India, and Nepal, Matroid is using computer vision to push the boundaries of Ophthalmology. We have created a high-performing model that learns to predict glaucoma from areas often ignored by doctors during diagnosis - specifically Lamina Cribrosa, as no established metrics exists yet for this region. We were able to detect glaucoma on OCT scans of the eye, with an F1 score of 96% and similar AUC and accuracy.

Obfuscate your digital presence with Matroid's intuitive API.

We are excited to announce that Matroid has been named a “Cool Vendor” by Gartner, Inc.

We are proud to announce our partnership with Dell to run Matroid On-Premise on Dell machines.

It’s the worst part of making the best detectors: meticulously curating a robust set of training images that prepares your detector to recognize the things you care about in the real world. Unfortunately it’s also the most important part of creating detector; if it’s garbage in, it’ll be garbage out, regardless of how state of the art your model is. Luckily, Matroid has a host of features that make training image curation as painless as possible.

We are proud to announce our partnership with HP to run Matroid On-Premise on HP machines.

Cheap storage, ubiquitous connected cameras, and an explosion of user generated content have led to an exponential increase in the amount of visual media on the internet and company servers. Unfortunately, the tools to explore this content have lagged behind: the images and videos are there, but how do you find what you’re looking for?

This post originally appeared on the Oracle blog and is reproduced here with permission:  Part 1, Part 2, Part 3

What do a patent application drawing for troll socks, a cartoon scorpion wearing a hard hat, and a comic about cat parkour have in common? They were all reportedly flagged by Tumblr this week after the microblogging platform announced that it would no longer allow NSFW content. Wired discusses the drawbacks of using Computer Vision without exports to help.

Two Turing Award Winners, the creators of TensorFlow, PyTorch, Spark, Caffe, TensorRT, OpenAI, and others will lead discussions about running and scaling machine learning algorithms on a variety of computing platforms, such as GPUs, CPUs, TPUs, & the nascent AI chip industry.

Computers are now excellent at recognizing images of human faces, cats and dogs — but struggle when it comes to detecting continuous actions, for example determining if a character in a video might be “dancing the tango.” Computers also fall short in detecting nuanced expressions of human emotions.

Billy Dally's talk transcript at ScaledML 2018, with video and slides, speaking about Scaling ML with NVIDIA GPUs.

Matroid's logo was drawn beautifully by a local artists in Beijing China, at The Summer Palace.

Ilya Sutskever's talk transcript at ScaledML 2018, with slides and video

Anima Anandkumar's talk, video, slides, and transcript at ScaledML2018

Jeff Dean's talk, video, slides, and transcript at ScaledML2018

The "TensorFlow for Deep Learning" Book has been released with code examples on the Matroid Github.

We had a fantastic gathering of 700 researchers and engineers, all learning about Scaling ML along different dimensions. Slides, videos, and photos of the event are included below.

Join us & the creators of TensorFlow, Caffe, Keras, Spark, CUDA, MapReduce, OpenAI, & others, giving talks at The Scaled Machine Learning Conference 2018.

Where: CEMEX Auditorium, Stanford University
When: March 24th, 8:45am - 6pm
Details: http://scaledml.org

An overview of the internals of Matroid presented at the Amazon Web Services annual conference.

A gigantic shift in computing is about to dawn upon us, one that is as significant as only two other moments in computing history. First came the “desktop era” of computing, powered by central processing units (CPUs), followed by the “mobile era” of computing, powered by more power-efficient mobile processors. Now, there is a new computing stack that is moving all of software with it, fueled by artificial intelligence (AI) and chips specifically designed to accommodate its grueling computations.

Matroid has raised $10 million Series A co-led by Intel and NEA, bringing funding to $13.5M, following a $3.5M seed round from NEA. We are excited to work with Intel and NEA to continue bringing Computer Vision to everyone. Join us! We are hiring at matroid.com/careers

Today the Internet Archive announced "Face-o-matic", to identify prominent public figures in TV channels aired across the world.

Matroid has been covered by The Wall Street Journal, Bloomberg, MIT Technology Review, TechCrunch, BBC, Seattle Times, Register, CNBC, and many others.

Matroid was presented at the Fermilab special colloquium on May 9th. Slides and a video of the talk are available on the Fermilab lab colloquium page.

First Two Chapters of "TensorFlow for Deep Learning" Released, head over to the O'Reilly website to download your copy. A PDF of the first two chapters is also available exlusively via Matroid here.

Matroid launched at Scaled ML 2017, and was covered by The Wall Street Journal, Bloomberg, TechCruch.

The Scaled Machine Learning 2017 Conference Slides.

Come see what we've been up to, learn about leading deep learning cloud providers, and meet leading researchers in the area! More details at http://scaledml.org

In this post, I explore the “hardness” in optimizing neural networks and see what the theory has to say. In a nutshell: the deeper the network becomes, the harder the optimization problem becomes.

As Intel moves aggressively into Machine Learning, they have published an interview with Matroid.

The FusionNet architecture will be presented at the NIPS 2016 workshop for 3D Deep Learning.

Organized by Amazon and Matroid, the DEEM workshop will be held in conjunction with SIGMOD/PODS 2017 in Raleigh, North Carolina. DEEM aims to bring together researchers and practitioners at the intersection of applied machine learning, data management and systems research, with the goal to discuss the arising data management issues in ML application scenarios.

Leading a group of industry and academic contributors, Matroid receives the KDD 2016 Best Paper Award runner-up in the applied data science track.

Our new Neural Network architecture named FusionNet is leading in the Princeton ModelNet challenge.

"That openness, and continual Google updates, have lured developers like computer-vision startup Matroid, which re-wrote its software to work with TensorFlow, after building on another free AI tool called Caffe."

Slides from the Matroid Scaled Machine Learning Conference held on Stanford campus.

We are excited for the upcoming Scaled Machine Learning Conference on Stanford Campus. See the lineup at http://scaledml.org. At Matroid we have been using TensorFlow extensively, and to help the Open Source community learn more about this tool, we are running two sessions devoted to it at our company conference.

Announcing the Scaled Machine Learning Conference on Stanford University Campus, Matroid's company conference.

Several weeks ago, Innovation Endeavors partnered with Bloomberg BETA to host an exclusive deep learning event for top experts – academics, entrepreneurs, and engineers – and investors to take a closer look at the space. The event centered around a panel discussion with Ilya Sutskever, Research Director at OpenAI; Reza Zadeh, CEO at Matroid and Professor at Stanford University; and Richard Socher, CEO of MetaMind. The conversation was moderated by Jack Clack of Bloomberg.

Today O'Reilly released a report on the Future of Machine Intelligence. The concluding chapter is an interview with Matroid.

Join us at Strata 2016 in San Jose, on Thursday March 31st, in room 210 A/E. Reza from Matroid will be hosting the Spark session. See here for a full schedule.

Matroid recently moved into a new office! Visit us at 3239 El Camino Real, Palo Alto 94306.

Located next to Caltrain and the restaurants of California Avenue. 

How we chose the name "Matroid", a concept from Computer Science Theory and Mathematics.

We were at the Strata + Hadoop world conference this past week and presented in the "Hard Core Data Science" track.