Megh Video Analytics Solution Use Cases

The explosion of real-time data collected from countless devices and sensors drives a seemingly unquenchable demand for edge computing and real-time analytics. To help organizations keep up, advances in artificial intelligence (AI) are enabling them to extract ever more sophisticated levels of appropriate information from streaming data. What's more, intelligent video analytics can now support a wealth of new use cases beyond essential security surveillance. In short, AI-powered video analytics is proving to be a killer app for the edge – and if you're not yet investing in edge computing, you're in significant danger of falling behind the curve.

Previous video analytics solutions at the edge have been constrained by performance challenges that delayed time to insight; as a result, companies have had to settle for a reactive approach to data when they need to be proactive for real-time decision making. WWT technology partner Megh Computing has addressed those challenges and answered the need for faster insight with its new, scalable Video Analytics Solution (VAS) portfolio powered by Intel® technology. And, it's opening up all-new use cases and possibilities for edge to cloud computing analytics. 

To showcase those new capabilities, WWT has built a VAS environment in its Advanced Technology Center (ATC) that allows customers to explore this solution applied in a variety of use cases, including those that replicate their own company's environment and needs. 

A quick look at video analytics, then and now

For those unfamiliar with this technology, intelligent video analytics is the ability to derive actionable insights from the content of video frames, detecting such aberrations as unusual motions, objects out of place, intrusions into restricted spaces, as well as crowd movement patterns.

If motion detection itself seems a little underwhelming – those sensors have been around for decades – the modern twist is the AI-powered ability to discern significant or unusual elements in the video streams and recommend courses of action. Previously data could only be captured, stored and later analyzed (maybe) by human eyes. This method of analytics was reactive and limited in its ability to provide actionable information quickly. It's been reported that more than 95 percent of recorded video footage is never even reviewed.

Earlier analytics technology was further constrained by the performance capabilities of hardware and software to model data accurately or to scale up and out affordably. But, those days of lagging and limited analytics are over: Megh Computing has introduced a new generation of scalable real-time intelligent video analytics solutions with high-performing, low-latency AI acceleration, applicable in a diversity of use cases from retail and manufacturing to logistics, security and more. Megh's real-time analytics platform leverages the latest Intel® technologies such as the Intel® Distribution of the OpenVINO™ toolkit to speed up AI workloads, with the flexibility to introduce and expand any use case – start small and scale for growth as needed.

Megh's VAS portfolio offers all-new levels of performance, flexibility and scalability 

As part of our commitment to working with the world's leading innovators in a global technology ecosystem, WWT forges alliances with longstanding partners such as Intel and emerging stars like Megh Computing. Founded in 2017, Megh has already established a solid alignment with Intel's approach to deriving maximum performance from hardware and software working together.

For its part, Intel invests heavily in nurturing the most promising software solutions, recognizing the importance of software/hardware integration for optimum performance and reliability. This imperative for world-class collaboration is central to how Megh and Intel came together to develop a solution that leverages Intel's vast portfolio of products and technologies. Megh's video analytics platform runs on flexible, scalable Intel technologies that support open-source environments from edge to cloud.

For example, the Intel Distribution of OpenVINO toolkit for integration with open source frameworks enables organizations to optimize, tune, and run comprehensive AI inference with write-once, deploy-anywhere efficiency. Intel® processors bring AI-assisted acceleration to every VAS environment, ranging from Intel® Core™ processors for desktops and laptops to the latest 3rd Generation Intel® Xeon® Scalable processors which are optimized for a range of workloads and performance levels and offer unparalleled flexibility. In addition, Intel® Stratix® Series FPGAs accelerate VAS by combining high density and performance with a rich feature set to maximize functionality and system bandwidth.

Together, these technologies contribute to a hardware/software solution stack that can flex and scale by use case. 

From edge to cloud and back, Megh's VAS use cases are limited only by imagination

Megh's VAS is built using its Nimble application framework to deliver significant flexibility and scalability, as well as high performance. Nimble implements the complete video analytics pipeline and can be deployed using either the CPU only or with FPGA accelerators. 'It's easily customized by modifying the pipeline with new deep learning models and analytics libraries, and by integrating other applications.

For the performance needed to process video analytics in (near) real-time at the edge, Megh's solution can ingest multiple channels of HD video. As video data is captured, users can apply predictive and prescriptive analytics to the data as the system intelligently discerns what's happening – extending the possibilities of motion and object detection, intrusion alerts, traffic management and more. And, it can suggest courses of action. The insights from this streaming data can help organizations accomplish crucial objectives as to enhance safety through social distancing, body temperature and mask compliance; control access to restricted areas via smart surveillance; aid inventory management to decrease misplacement; and help prevent fraud, theft and other unwanted behavior.

Megh's VAS offering can scale up and out according to customer needs. For example, 10 or fewer cameras might be appropriate for specific edge gateway applications, while up to 50 cameras could be employed for edge servers powered by Intel® Xeon® processors. The solution scales to support 150 cameras and more. 

Here are just a few of the use cases this platform supports:

• Museum exhibits automatically activate a warning alert when visitors try to walk too close or touch them.
• Potential fire hazard warnings when too many people congregate in an area with limited egress.
• Stadium foot traffic analysis to determine strategic pillar placement that facilitates efficient foot traffic.
• Retail traffic monitoring for optimum kiosk placement, product layout and so forth for richer in-store customer experiences.
• Loitering prevention to detect and discourage potential troublemakers gathering near storefronts or parked vehicles.
• Collision detection, a new use case with the latest release, detects incidents and notify safety authorities in industrial and smart city environments.
• Monitoring industrial output to assess the most cost-effective variable insurance rates, depending on moment-moment productivity – for example, whether an oil field is pumping petroleum or mud.
• Schools can rely on video analytics to promote playground safety, distinguish intruders from staff, monitor hallway traffic, and improve fire drill response times.
• Manufacturing facilities can enforce safety by ensuring that employees and visitors stay clear of hazardous areas.

WWT showcases Megh's video analytics capabilities in the ATC

Edge computing is an ascendant technology, and video analytics offers numerous advantages to make critical decisions with confidence based on near-real-time insights. But, it requires the right compute hardware and software working together to deliver powerful performance, flexibility and scalability. 

Nearly every edge analytics solution uses AI (personas), rule-based machine learning models and policies.  Data is either ingested into a cloud solution or moved on-prem, where model training happens; it's then pushed to the edge solution. Megh is changing this process by training a model locally at the edge of the VAS solution. The result is an AI-driven video analytics platform that leverages Intel technologies and software for a complete solution. Compared to cloud-deployed solutions that could take up to three times longer to obtain an actionable response, Megh's VAS solution allows customers to deploy analytics at the edge for dramatically faster results.

The ATC has already built out multiple video analytics use cases to demo the solutions in action. WWT can even build out customers' specific use cases in the ATC to illustrate their effectiveness, flexibility and scalability. And, we can show you the enabling Intel technologies that drive Megh Computing's real-time livestream video analytics platform.