CIO Insider

CIOInsider India Magazine


VisionBot: Deriving Priceless Business Insights from Visual Data Store

Amit Chakraborty & Santhosh Nair,Co-Founders

Thanks to cutting-edge optical technologies, the world is flooded with visual content. The retail businesses on a daily basis generate tons of video footage and images, which are often only used for security applications. VisionBot is revolutionizing the retail industry by giving priceless business sense to the visual data junk. It helps businesses analyze their visual content to derive powerful insights and trends.

A team of developers and programers with backgrounds in data science, machine learning and cloud services, VisionBot is able to do what most humans cannot easily do - observe a trail of video frames 24X7 and derive insights. The company is in a constant pursuit of developing new solutions around AI in Computer Vision to help businesses optimize their potential by using their own otherwise-unused visual content. CIO Insights engages in an exclusive interaction with the company’s co-founder Amit Chakraborty to know more about its exciting endeavors.

In conversation with Amit Chakraborty, Co-Founder, VisionBot

The technologies like Computer Vision, Artificial Intelligence (AI) and Machine Learning (ML) are almost entering their fine-tuning phase. As a technology company at the cutting edge of things, what is the kind of challenge that VisionBot solves for its customers?
When it comes to traditional segments

ranging from manufacturing to logistics, education, and hospitals, the major wedge of their business happens offline, and AI today is starting to manifest its capability to improve their operational efficiency and customer satisfaction. But something that cripples this trend is the lack of technologies to acquire data that feeds the algorithms. It’s indeed no rocket science. Take the example of a hospital, wherein most of the activities involved are visual - right from patients entering the facility to leaving. Unlike in the digital world, the real-time data is generated in the form of subjective visuals and videos (like CCTV footage), and not in binary format, which makes it difficult to mine & analyze data. This is where VisionBot comes in. We help businesses to obtain objective data from subjective visuals. VisionBot is an evolution based on Neural Networks and Deep learning, and forms a culmination of the projects and development work we have done in the field of computer vision applications over the last few years.

We help businesses to obtain objective data from subjective visuals

When it comes to data generation, what are the parameters that we are talking about?
Take the example of a retail store. The only objective data they have is generated at the POS terminals, while their online counterpart incidentally records the number of visits, their demography, the items they searched for and what not. We equip a retail store to access exactly the same kind of information in an automated manner using the already installed (or by installing) surveillance infrastructure and powered by our computer vision and AI engines.

So, any business that wants to leverage the potential of the subjective visuals generated at their premises could be your customer. How do you equip

yourself to care for such a large spectrum?
While deciphering the logic, our systems are designed to see only two things — the object to be detected and the event to be detected. Regardless of whether it is traffic, a retail outlet, or a hospital, we enable them to detect the subject and do analysis in a real-time manner using our cloud platform. Thanks to the better latency we are able to provide, it doesn’t matter whether it’s a live stream or a recorded footage. For instance, let’s take the case of a zoom classroom, which has become a new norm during this pandemic. Using our engine, we are able to read the facial expressions of students during a class, and provide the school management with a meticulous analysis of each student, including the parameters like attention in class.

Could you tell us about one of your recent deployments?
A few months back, we deployed our solution at one of the mid-sized restaurant chains in the US. They were looking forward to having a quantitative estimate of customer satisfaction across their 20 to 25 restaurants. Interestingly, they wanted to decipher the data from cameras placed on the inside as well as the outside. In addition to information like the number of visits, the demography, and the facial expression of customers, we, additionally were able to automatically analyze & record the exact time taken by their staff to clean each table. We got kudos from them, as improving this aspect directly impacted their customer experience and therefore improved revenue generation.

Going forward, what are your future plans?
One of the predominant future plans is to make our systems more scalable. In addition, we also are looking forward to bringing down the price points - the setup costs as well as the monthly charges for each camera. Another aspect that we have our focus on is localization of our systems and, as a part of this, we also want to include voice analysis in our capabilities.

Current Issue
Cloud Migration In Manufacturing Industry