This post has been republished via RSS; it originally appeared at: New blog articles in Microsoft Tech Community.
Experimenting with technology and applying it to new scenarios expands our skills, feeds our ingenuity, and stretches the limits of what’s possible—something we’re always doing here at Microsoft. Today, the intelligent edge is transforming business across every industry. As we evolve from connected assets to connected environments and ultimately, to entire connected ecosystems, Microsoft continues to invest heavily to accelerate innovation of AI at the edge.
I’m particularly enthralled with Azure Percept, an end-to-end, low-code platform that streamlines the creation of highly secure artificial intelligence (AI) solutions at the edge. It works seamlessly with Azure Cognitive Services, Azure Machine Learning, and other services. The Azure Percept development kit includes pre-built AI models that can detect people, vehicles, general objects, and even products on a shelf. It also includes workflows for edge AI models and solutions. In fact, setting up a prototype takes mere minutes.
I recommend exploring Azure Percept to truly appreciate all the enticing possibilities. To inspire your creativity, I’ll explore some ingenious solutions that are driving value across manufacturing, retail, smart cities, smart buildings, and transportation. I’ll also share some step-by-step projects and other resources to help you get started with Azure Percept.
Building smarter cities
Cities are increasingly leveraging edge AI and IoT to improve safety and efficiency. Edge computing scenarios enabled by Azure Percept include:
- Conserving water and energy resources through utility management
- Optimizing transportation planning
- Improving citizen security by monitoring for suspicious activity
To keep people safer during the pandemic and improve city operations, Kortrijik in Belgium needed insights into pedestrian movement and density. The city turned to Proximus which used Azure Percept and machine learning to develop a proof of concept. Placing sensors in key locations, they were able to successfully deliver real-time vision and audio insights to promote safety and smoother operations.
The ability to monitor street activity and traffic can inform decisions around maintenance, road conditions, parking availability, and transportation planning. Using the Azure Percept dev kit, this enterprising developer was able to build a sample traffic monitoring application that could classify vehicles into categories, which could then be used to generate insights such as traffic density and vehicle type distribution over time. Connecting to LTE or 5G networks can also help Azure Percept enable smart city scenarios.
Making spaces safer and more efficient
The prospects for making homes, office spaces, and other facilities smarter and safer are seemingly limitless. Solutions built in Azure Percept can enable scenarios such as:
- Detecting "tailgating” by unauthorized individuals
- Monitoring for access control, intrusion detection, and suspicious activities
- Ensuring worker safety through dangerous zone monitoring and accident detection
Take, for example, real estate management companies, which like most businesses are always seeking opportunities to lower costs. By default, most office spaces turn up heat (or air conditioning) in the morning and turn it down in the evening, regardless of usage. This smart building business case for using Azure Percept shows how you can improve building climate control by tracking space usage and weather forecasts to optimize building temperatures—and potentially lower costs by 20%.
For those looking to get hands-on with implementing these types of smart spaces use cases, Microsoft AI MVP, Goran Vuksic, illustrates how the standard Azure Percept people detection model can be used to build a room occupancy detection solution. Other possible smart space application tutorials include how to control a desk lamp with voice commands.
The possibilities for retailers are especially appealing. Solutions created in Azure Percept can be used to reduce queue waits, aid in social distancing, and keep products in stock and on shelves. Additional use cases include:
- Understanding consumer traffic flows and shopping patterns
- Improving store safety and space utilization
- Expediting in-store pickup of online purchases through vehicle detection
For retailers that stock perishables, solutions built in Azure Percept can help reduce spoilage, a problem that leads to $161.6 billion in cumulative annual losses in the US.[i] For example, video AI can monitor bins of produce to send alerts when restocking is required. By applying insights over time, these analytics help store personnel understand how much to order. Purchasing analytics can also be used to increase supply chain efficiency, lowering emissions from the transport of fresh food, reducing use of fertilizer, and creating a healthier planet.
For those that are getting retail items delivered to their door, it’s also possible to set up your own end-to-end package delivery monitoring edge AI application using Azure Percept. Discover the exact steps needed to set up an AI application that detects when a person or truck is seen by a camera, all of which is then sent to a video analytics dashboard for real-time updates so you’re never left wondering if your package has been delivered yet.
Making manufacturing safer and more productive
For manufacturers, AI at the edge solves multiple challenges to create safer, more productive, and more profitable operations. Through intelligent automation, manufacturers can capture data insights throughout production and remotely monitor plants and warehouses. Azure Percept can be used to address a range of use cases to help manufacturers:
- Improve quality through defect detection
- Automate assembly processes and supply chain visibility
- Increase worker safety and loss prevention
As laid out in this industrial IoT business use case, AI edge solutions powered by Azure Percept enable preventative maintenance by detecting and diagnosing issues before they disrupt production. This saves both time and money. Furthermore, Azure Percept can even be used to control robotic equipment with voice commands to remotely exchange parts, adjust settings, and perform other service tasks.
Within manufacturing settings, there are often multiple people, vehicles, and other objects on a factory floor. Which is why being able to use the Azure Percept dev kit to create an obstacle avoidance model can be so valuable. Explore the steps taken to build the “Perceptmobile”, including how to train the custom vision model via Azure Percept Studio and test it with the camera stream.
Envision the possibilities for your industry or use case
These are just a few of the industries where Azure Percept can drive incredible value. Scenarios in healthcare include supply chain efficiency, patient recognition, and waiting room prioritization. Edge intelligence is being used in transportation to check and identify aircraft parts. It can even be used for improving people’s lives, like interpreting sign language.
Azure Percept makes creating value at the edge easier than ever. And the good news is that the Azure Percept DK and Azure Percept Audio are now available in 16 markets. To envision all the possibilities, I suggest that you dive in and start exploring Azure Percept today.
[i] USDA, ‘Loss-Adjusted Food Availability Documentation,’ available at: https://www.ers.usda.gov/data-products/food-availability-per-capita-data-system/loss-adjusted-food-availability-documentation/ (accessed Sept 7, 2021)