X

The latest cloud infrastructure announcements, technical solutions, and enterprise cloud insights.

Image Recognition Software Startup Takes on Big Players with Oracle Cloud Infrastructure

Mark Brunelli
Product Content Strategist

Image recognition software provider Netra is a fairly small player in the artificial intelligence (AI) market, but the company is using a high-performance, multicloud computing strategy to take on big players such as Google Cloud Vision and Amazon Rekognition.

Netra helps businesses make sense of the tsunami of digital imagery on the internet, said CEO and founder Richard Lee, who shared his company's story on stage at the O'Reilly Velocity Conference in San Jose, California.

Specifically, Netra uses computer vision, AI, and deep learning to help brands and agencies reach and better understand their ideal target audiences. The company's image recognition software analyzes billions of consumer photos to identify interests, life events, demographics, and brand preferences.

"We provide image recognition as a service to our customers, and we deliver that through an API that gives access to our deep learning models, which are trained up on over 7,500 classifiers today," Lee said. "So, this is a little bit more complex than Hot Dog or Not Hot Dog." For those that don't watch HBO's Silicon Valley, this refers to an app on the show that identifies whether an image is of a hot dog or not.

The deep learning models are deployed on Oracle Cloud Infrastructure and built on top of Apache Kafka, which is open source stream-processing software, plus Docker and Kubernetes. Netra's technology works by identifying objects of interest and looking for pattern matches around specific clusters of pixels.

"For example, our humans model may detect [a human face] and then send it to our humans daemon, which then classifies age, gender, and ethnicity," Lee explained. "Likewise, our brands model looks for the presence of a logo. … And then lastly, our context and object model detects and classifies what else is in the image."

The image recognition software accomplishes all this in about 200 milliseconds.

Why Oracle Cloud Infrastructure?

Netra's customer base has recently grown to include large enterprises, and with that comes a higher volume of images and videos to analyze, as well as higher service-level agreements. The Boston-based company is counting on Oracle Cloud Infrastructure to help it meet these increasing demands.

"Fundamentally, [Oracle Cloud Infrastructure] gives a startup like us access to machines that would cost us thousands to purchase on our own, as well as the flexibility to scale up and down as needed," Lee said. "Oracle gives us really strong value in terms of pricing and performance."

Lee said he likes the flexibility that Oracle Cloud Infrastructure provides, especially when there is a spike in demand for his company's services.

"If we get hit with a couple million images … we're able to spin up a new instance almost within minutes, to be able to work the queue down," Lee added. "Once the queue gets below a certain threshold, we're able to spin that down to manage our costs."

The deep learning models that Netra deploys in the Oracle cloud are very complex, and the amount of compute power it takes to process photo and video is "pretty intense," Lee said. 

"We are always waiting for the next-generation GPU chips to be released," he said. "We're constantly pushing the envelope on the processing side, and we're always looking for the highest-performance hardware available. And from what we've seen, Oracle Cloud Infrastructure is the best price/performance on the bare metal side so far."

Oracle gives startups such as Netra the computing horsepower necessary to train up deep learning models and compete with some of the biggest players out there. Running AI models in the cloud also gives Netra more bandwidth to focus on its core value proposition. 

"With Oracle Cloud Infrastructure, it's not a matter of how big your capital budget is, because it's kind of democratized for everybody," Lee said. "Now it's more about: How good are your computer vision models? What kind of solutions can they build? In that case, it's a much fairer fight against competitors, and we're excited to be able to participate. That would have been impossible before the advent of cloud and really the cost/performance that Oracle Cloud Infrastructure has provided to us."

Accelerating to the Cloud

Netra also takes part in the Oracle Cloud Startup Accelerator program, which helps startups get up and running in a short period of time.

Program participants can take advantage of several benefits, including free credits for Oracle Cloud Infrastructure, world-class mentoring and consulting, start-of-the-art cloud technology, coworking spaces, and access to Oracle customers and partners.

Lee especially likes the fact that his company can now get noticed by hundreds of thousands of Oracle customers—and the free credits certainly don't hurt.

"It's like nondilutive venture capital," he said.

Sage Advice

Lee advised that other startups that are considering a move to an enterprise cloud platform should take advantage of those free credits that cloud providers offer.

"There is a lot of money to get started and build apps and to actually run high-performance services that are effectively free funding right now," he said. "So as a startup, you can really extend your runway with these credits. But in order to do that, you have to be smart about your architecture and how you deploy it. For example, we've used Docker containers or Kubernetes to be agile to be able to deploy across multiple providers and services."

And don't forget to look for the best solutions in terms of pricing and performance.

"I think it's an amazing time to start a company," he said. "You need fewer resources than ever before, and you can scale faster than ever before through a lot of these startup-type programs."

Be the first to comment

Comments ( 0 )
Please enter your name.Please provide a valid email address.Please enter a comment.CAPTCHA challenge response provided was incorrect. Please try again.Captcha