Stock Analysis, IPO, Mutual Funds, Bonds & More

Why the 'Father of the Pentium Chip' is bullish on image processing startup Orbo.ai

In the post compression era and with new improved hardware launching every month, video processing continued to remain stuck in low resolution status to save the bandwidth. Orbo looks to change that.

, ET Online|
Updated: Dec 11, 2019, 12.28 PM IST
Recently the company raised $1.6 million in Pre-Series A funding, which also saw participation from Vinod Dham, developer of Intel’s Pentium microprocessor. However it took time for investors to come around.
Hot Startups 2020
Founded in 2015, Orbo started on the premise that there has been a consistent amount of low quality and low resolution imaging that exists in multiple business segments, from surveillance to medical images to ecommerce images.

“These impact the customer delight, the conversion, the delivery and the ecosystem or the hardware that they are related to. We wanted to go beyond the legacy image processing and the limitations of the hardware, where we could bring in AI into the mix and that kind of a customer delight which, at the end of the day, would solve all these indirect problems of conversions,” explains Manoj Shinde, Founder, Orbo.ai.

Shinde reiterates his point by putting forth the example of ecommerce giant Amazon, which can lose a whopping $1.6 billion per second if the images are not handled properly or if the latency of the images on the ecommerce sites are not handled. “If something doesn't download, well, a customer has already moved from one place to another looking for the next product,” he shrugs.

Building from scratch
The founders saw that in the post compression era and with new improved hardware being launched every month, video processing continued to remain stuck in low resolution status to save the bandwidth. This created a big gap between the device’s resolution versus the one streamed or processed by the server.

“We wanted to create the imaging experience as seamless and qualitative as possible for billions who were stuck in the technology limitations and wanted to create an amazing domino effect where network providers could save extra bandwidth, content providers saved cost on creations and customer got the best experience as well as enhanced resolutions,” reflects Shinde.

However, though Shinde and his co-founders Abhit Sinha and Danish Jamil were able to identify the core issue, they soon realised that it would not be an easy ride. “We had every intention to go beyond the limitations of the existing camera ecosystem. But, in 2015, we didn't have the ecosystem, the frameworks and neither any network that could help us to build the capability of AI research. So we built a lot of the components from scratch and hence became self reliant over time,” he recalls.
The company spent almost three years in AI research before they started commercialising. Since then Orbo has built a significant IP in image and video enhancements that has been a pull for many business segments where video resolutions are a problem. “Unless you are under the umbrella of one of the blue chip tech companies like Adobe, Google etc, it can be very difficult and incredibly expensive to start something that is driven by research, especially to run deep learning or computer vision futuristic experiments for image enhancements,” he avers.

Currently, Orbo offers three sets of business solutions - Super Resolution, which claims to upscale any low resolution visual to high resolution without compromising the quality, aesthetics and performance. A subset of this is SmartFix, an AI-based ‘one-click’ image enhancement framework that can solve problems for ecommerce firms, content creators and hardware manufacturers by going beyond physical pixels.

In their Deep Facial Model, the company has trained neural networks using thousands of faces to identify landmarks such as jawline, eyes, nose, lips, and have created a deep facial model that can operate in mobiles phones in either 2D or 3D environment. Beauty companies, expectedly, are utilising such tech to enhance consumer facing apps for facial makeovers. Lastly, their AI research has produced more than 16 computer vision and AI-based IP driven components for image transformation or enhancement.

The India opportunity
As far as the Indian market is concerned, Shinde is of the view that India is a bandwidth strapped country. “Although the prices are very low, but the penetration of towers, network congestion and quality hinders the performance for the users,” he says.

He cites the example of education video content providers who would find it incredibly difficult to bring in the next set of million users from tier 2 and 3 cities since 4G isn’t easily prevalent in such cities. The content platforms will end up streaming low resolution content which, in essence, would deter user conversion. “Such use cases are in abundance and our product being executed on the edge (meaning software running on the device instead of servers) will enhance the quality even if they are fed with low resolution. This applies to all Over-the-Top (OTT) as well as any Video On-Demand (VOD) platforms. It will also help struggling telecom operators to optimise their backend haul systems as 5G starts rolling out,” adds Shinde.

5G world video streams, he feels, will cost a fortune to move from core to edge and in the VOD world, 90% of the content is still standard definition. Orbo’s technology can come in handy and be embedded in software or in silicon, designed to enhance pictures, documents and videos. Some of the popular use cases include enhancing low light in smartphone images, removing the fog from security camera footage or improving the resolution for movie theatre transmission.

Orbo founders: Manoj Shinde, Abhit Sinha and Danish Jamil

According to Sudin Baroakar, a Global IT & Innovation Advisor, in the overall AI space, computer vision and ML behind it has a huge potential in India. “The capabilities where you add an AI layer and take an appropriate decision has a great scope and this space will continuously rise. Use cases can be seen in sectors such as healthcare, defence, security, ecommerce and digital banking products,” says Baroakar. He adds that Deep Tech is the only way to go up the curve. “A combination of product and deep tech stack is the best way forward for the Indian startup ecosystem to evolve,” he says.

An evolving model
Media and entertainment, understandably, is a top priority for the company. So is the ecommerce segment. Now Orbo is also paving its path in the healthcare and 5G segments. “These are evolving segments for us. They are large and very scalable. Think about 5G - the footprint is going to be monumental. In fact, our hypothesis and research are getting major endorsements in the form of partnerships such as Intel, Samsung from the device standpoint, Cisco from infrastructure and even 5G operators such as Vodafone and Verizon,” he reveals.

Although the company had earlier stayed away from healthcare due to the regulation and compliances aspect, it is now ready to step in. Shinde highlights how Orbo can be revolutionary for the healthcare segment. Substantiating his argument, he talks about the CT scan images where multiple hops are run to create multiple slices. So, for instance, in order to create 30 slices of CT scan, it may take 60 seconds of iterations of hops on the scan. This, in turn, implies that the quantum of time one stays inside the machine and the amount of radiation that one is exposed to will be of a longer duration. “They do multiple hops to get the exact clarity of that image or the resolution of that image so that they don't lose out on any particular detail. We are coming in halfway - reducing 60 seconds to 30 and instead of 30 slices take 15 and build a super resolution imagery on top of it which will help the patient get exposed to lesser radiation,” says Shinde, the enthusiasm palpable in his voice.

Faith comes to the fore
Recently the company raised $1.6 million in Pre-Series A funding, which also saw participation from Vinod Dham, developer of Intel’s Pentium microprocessor. However it took time for investors to come around. “The struggle that I had initially was to showcase what we are doing and how valuable it is. It was really difficult for investors to really understand the landscape of this product in the beginning. Now everybody brings up the use cases with us and everybody is willing to invest in us,” says Shinde.

Moreover, when they were thinking about Pre-Series A in 2017-18, deep tech was still not resurfacing as the prime segment. “We were a believer of thematic investment in India - we have seen food tech, fintech, blockchain, it is sort of like a consortium that comes in and then that defines the area to invest in. Fortunately, 2019 is the area of deep tech which everybody is vying to get into,” he chuckles.

Run a business? Take part in India's definitive ranking for Micro, Small & Medium Enterprises

Also Read

Intel’s Pentium microprocessor developer,Vinod Dham, others back Orbo AI

Add Your Comments
Commenting feature is disabled in your country/region.

Other useful Links

Copyright © 2020 Bennett, Coleman & Co. Ltd. All rights reserved. For reprint rights: Times Syndication Service