Podcast #4: Understanding AWS Panorama and edge computing

General Manager of AWS Panorama, Omar Zarka, reveals the benefits and uses of this AWS edge computing service.

— by Miguel Bracchini, Chief Solution Architect -

In episode four of the AWS Insiders podcast, AWS superfan and CTO of ESW Capital, Rahul Subramaniam, interviewed General Manager of AWS Panorama, Omar Zarka, who provided an in-depth look into the structure of AWS Panorama, a machine learning appliance and software development kit that brings computer vision to on-premise internet protocol cameras. Listen to the full podcast episode here, or read on to see how Amazon Panorama can make accurate predictions, and learn how to reduce operational overhead and improve customer experience.

The genesis of AWS Panorama

When developing Panorama, AWS began by trying to solve an education problem. As AI and machine learning became more prevalent in the cloud, a gap emerged in what developers were able to do, and the skills they needed to take advantage of these new technologies. AWS’s solution was to create programs to educate people on new concepts so that they could see how to apply different services to their particular business problems. One of the first initiatives was called Deep Lens – a fully packed camera with edge compute, which allowed you to run machine learning models.

The idea was to give people a platform to learn what it meant to run deep learning models on computer vision use cases. Its popularity grew and multiple interesting use cases emerged, including sign language interpretation, and homework help for children. Beyond these creative applications, businesses also began buying these cameras in large quantities and tried to deploy them in their own processes. Intrigued by these buying patterns, AWS reached out to those customers to see what they were trying to achieve. They soon discovered that customers loved the simplicity of Deep Lens – an all-in-one package that allowed easy deployment of machine learning or computer vision to the edge. From this insight, AWS built a mini server-like smart appliance that attaches to your network and existing cameras and simplifies deployment of visual inspection. This is where AWS Panorama began.

AWS Panorama and edge computing

AWS has a variety of solutions that try to solve problems around edge to hybrid, edge to cloud, or cloud to edge, such as the AWS Snow Family and AWS Outposts Family. For Panorama, AWS saw that customers had use cases which needed the power of AI in the cloud, but specifically at the edge with the cloud’s scalability.

Omar identified to following three barriers that customers face when considering Amazon Panorama:

  1. Cost – the bandwidth cost of streaming camera data to the cloud is huge.
  2. Latency and reliability – some companies require a sub-100 millisecond latency for events, such as worker safety or emergencies, while other companies will lose millions if there is any downtime.
  3. Data regulation – many companies want more control over where their data goes, as well as the ability to discard data immediately after processing it.

When it comes to cost, Omar recommends leveraging the cloud first – you don’t have to buy hardware, it’s not a capital expense, your ability to scale your elasticity is high, and you can control costs daily instead of planning them over long periods. But some use cases are more suited for the edge, and are simply not solvable right now in the cloud. This is where you can leverage services like AWS Panorama.

AWS Panorama use case: Deloitte

One of the unexpected Amazon Panorama use cases that Omar presents is AI for animals that Deloitte in the Netherlands created to detect animal abuse in real-time. For Deloitte, it wasn’t just about identifying when the abuse happened, but it was also about intervening to minimize the abuse. This edge computing use case combines the advantages of edge AI in a very important, positive way, and was driven by three factors:

  1. It is difficult to have people monitoring animals all the time.
  2. Humans are a big part of the problem.
  3. It is often hard to respond in real-time, or in locations where the bandwidth or infrastructure necessary for streaming video to the cloud is limited.

AWS Panorama best practices

For anyone considering deploying AWS Panorama, Omar provided the following advice:

  1. Engage with a practitioner or AWS partner that understands how computer vision works at the edge, and what can be enabled. There is a lot of brainstorming that must occur around your specific use case. The core for every use case is similar, but the actual implementation and the differentiators for your particular situation can vary greatly. The scalability and flexibility of the AWS cloud is hugely beneficial, but that same scalability can be complicated at the edge.
  2. Think about why edge computing is right for your business. If you go down this route you absolutely can see improvements in your processes and results, but it must be the right fit for your business. So be precise about why the edge matters in your use case, and consult AWS to help you figure this out.

Additional insights from this AWS Panorama discussion between between Rahul Subramaniam and Omar Zarka include:

  • How to set up AWS Panorama
  • Common AWS Panorama models or analysis
  • Industrial applications
  • Healthcare and sports applications
  • Cost optimization recommendations

Listen to the full podcast episode here, and visit our podcast and videos page to listen to our full list of AWS Insiders podcast episodes.

More Updates

Cost management

CloudFix now supports Financial Savings

Since the launch of CloudFix cost optimization platform from DevGraph we have seen huge interest and adoption. Today, we are excited to announce that you can save even more with our new financial savings capabilities in addition to resource related

Read More
Skip to content