Can you differentiate between a definite buyer, a potential buyer, and a window shopper just by looking at their CCTV footage? Well, it may now be possible with video analytics, which can recognize spatial and temporal data of an object and analyze its actions or particular activities based on its movements. Deep North caters to this segment of the analytics industry by facilitating companies to implement real-time insights through video analytics. Analytics Insight has engaged in an exclusive interview with Azhaan Merchant, SVP of Strategy & Business Development, Deep North.

1. With what mission and objectives, the company was set up? In short, tell us about your journey since the inception of the company?

We exist because we believe data and deep learning can have a humanizing effect on real-world experiences. Our mission is to empower enterprises that operate in the physical world with accurate, real-time, decision-making insights so that they can create richer real-life experiences and better business outcomes.

Deep North was founded in 2016 by Rohan Sanil and Jinjun Wang, an expert in multimedia signal processing, pattern recognition, computer vision, and analytics. In 2016, Sanil pioneered object detection technology to help drive targeted advertising from online videos. When a major brand saw this, they challenged him to create a means of identifying, analysing, and sorting objects captured on their security video cameras in their theme parks. This exploration inspired development that would unlock the potential of installed CCTV and security video cameras within the customer's physical environment and apply object detection and analysis in any form of video. Deep North was then born to fully realize the transformative potential of AI in the physical world so businesses can deliver better, safer, and more exceptional customer experiences.

After rebranding in 2018, Deep North expanded the availability of its computer vision and video analytics products, to a variety of markets including retailers, grocers, airports, shopping malls, restaurants, and events.

2. Tell us how your company is contributing to the AI/Big Data Analytics /Cloud Computing industry of the nation and how the company is benefiting the clients.

We are pushing the frontiers in intelligent video analytics and have partnered with tech giants such as Nvidia and Dell to ingest massive amounts of video data and run our proprietary computer vision algorithms using both on-premise and cloud-based infrastructure. Our systems architecture has OCI-compliant containers running on VMWare with GPU virtualization on top of Dell edge servers. In addition, we have recently migrated to NVIDIA's Ampere microarchitecture which will allow us to increase the volume of cameras we can use for each GPU card available.

All of the metadata from the video feeds is sent to the cloud where our inference pipeline utilises our library of over 114 million objects to yield unparalleled accuracy in algorithms and analytics. This data is then visualised in real-time and can be viewed on Deep North's web portal, or mobile application or it can be pushed directly to any business intelligence tool. With less than a second of latency, we are able to provide real-time alerts directly into our client's dashboards or instant communication headsets so they can adjust store operations in real-time.

3. What is your biggest USP that differentiates the company from competitors?

A majority of the existing video analytics vendors use facial recognition to track a consumer's behaviour across physical environments – however, with privacy laws like GDPR and CCPA it is not possible to utilise such software without the consent of the consumer as it is stores personally identifiable information (PII). As a result, a majority of vendors deployed in the West focus on analysing and identifying objects from a single point of views such as the entrance doorway or check-out counter.

Deep North's main USP is its reidentification patent which uses skeletal-based tracking algorithms that divide each individual skeletal into 124 different vectors. The combination of these vectors similar to a fingerprint is unique to each individual and is then stored as an anonymized hash code in our cloud database. Due to this algorithm, we are able to track individuals across multiple cameras as we can re-ID each skeletal with their anonymised hash code when they move from one camera angle to another – which in turn permits us to stitch together a customer's journey from entry to exit in a completely GDPR compliant manner without storing a PII. This level of multi-layered data is much more useful for enterprises as they can get understand what the entire customer journey looks like for various cohorts.

4. How is your company helping customers deliver relevant business outcomes through the adoption of the company's technology innovations?

Currently, most operators only have access to point-of-sale data within their brick & mortar stores and there is an entire black hole of information on what a customer does from the time they enter the store to the time they exit. Deep North helps operators unlock several new tranches of data relating to what the customer is doing at every point- What demographics are predominantly coming into the store? Where exactly did they spend time? Which products did they touch? Was there a sales associate present to help them? Was there a long line at the check-out? Getting access to this information allows operators to understand the customer funnel more clearly so they pro-actively improve the in-store experience for each of their stores.

After this data is unlocked, Deep North's algorithm optimizes perpetually over time and its predictive analytics engine pro-actively guides enterprises on how they should run their stores. Invaluable suggestions are provided such as – What is the ideal store layout? How many labor hours are required in the front and back-off house? Where should you place your sales associates? What is the ideal product mix relating to the customer demographics entering that store?

5. How do you see the company and the industry in the future ahead?

As companies like NVIDIA continue to spend billions on advancing their GPU microarchitecture, the efficiency and computational capacity of deploying computer vision-based platforms are improving exponentially each year. In addition, as 5G network infrastructure continues to roll out at scale, it becomes more practical to use off-premise architecture as you can upload the video data directly from the cameras to be processed in real-time in the cloud. Both of these advancements on their own will result in increased adoption of video analytics solutions like Deep North as operators can light their stores up overnight with minimal costs and hassle. Once the cost and time to deploy there is little reason for brick & mortar store operators to not unlock the massive tranches of data sitting in their video assets – it will soon become as essential as Google Analytics is for e-commerce.

6. Which industry verticals are you currently focusing on? And what is your go-to-market strategy for the same?

A majority of our revenue today comes from big box retail, supermarkets, and shopping centres. We are also deployed with some airports, QSRs, and warehouses. Our go-to-market strategy is to find an anchor Global 2,000 customer in each of sector who has a voracious appetite to ingest new pockets of data as an accelerant to change the way they run their stores. We work with these anchor clients to understand what business outcomes our video analytics solution can help them solve and once we understand their goals, we can re-purpose our core computer vision algorithms to help them unlock data from their existing video assets in their brick & mortar stores. After identifying the immediate pathways towards ROI, we also create a road map of unique algorithms and features that we can develop to further enhance their in-store experience. After successfully validating our thesis across multiple stores and formats, we then roll out a blueprint across their entire format. We can then build a case study and white paper that will help us approach similar clients in different geographies who are mostly facing the same problems.