At Microsoft Build 2021, we are pleased to announce the availability of https://azure.microsoft.com/products/video-analyzer as an https://azure.microsoft.com/product-categories/applied-ai-services/. Azure Video Analyzer (AVA) is an evolution of https://azure.microsoft.com/services/media-services/live-video-analytics/ (LVA) that was launched several months ago. As an Azure Applied AI service, AVA provides a platform for solution builders to build human centric semi-autonomous business environments with video analytics. AVA enables businesses to reduce the cost of their business operations using existing video cameras that are already deployed in their business environments.
Many businesses take a manual approach to monitoring mundane business operations. This is prone to a higher degree of error and miss-rate due to human error and the fact that assessing the criticality of a visual event sometimes requires corelating multiple signals in real-time. AVA, in conjunction with other Azure services, addresses this challenge by providing a mechanism for businesses to automate mundane visual observations of operations and enable employees to focus on the critical tasks at hand.
AVA is a hybrid service spanning the edge and the cloud. The edge component of AVA is available via the https://aka.ms/ava-marketplace and is referred to as “Azure Video Analyzer Edge,” and it can be used on any X64 or ARM 64 Linux device. AVA Edge enables capturing live video from any https://en.wikipedia.org/wiki/Real_Time_Streaming_Protocol enabled video sensor, analysis of live video using AI of choice, and publishing of results to edge or cloud applications. AVA Cloud Service is a managed cloud platform that offers REST APIs for secure management of AVA Edge module and video management functionality. This provides flexibility to build video analytics enabled IoT applications that can be operated locally from within a business environment as well as from a centralized remote corporate location.
AVA can be used with Microsoft provided AI services such as Cognitive Service https://docs.microsoft.com/azure/azure-video-analyzer/video-analyzer-docs/analyze-live-video-custom-vision and https://docs.microsoft.com/azure/azure-video-analyzer/video-analyzer-docs/computer-vision-for-spatial-analysis as well as AI services provided by companies such as https://docs.microsoft.com/azure/azure-video-analyzer/video-analyzer-docs/use-intel-grpc-video-analytics-serving-tutorial. It is also possible to build and integrate a custom AI service that incorporates https://docs.microsoft.com/azure/azure-video-analyzer/video-analyzer-docs/analyze-live-video-use-your-model-grpc custom AI models. Business logic that corelates AI insights from AVA with signals from other IoT sensors can be integrated to drive custom business workflows.
AVA has been used by companies such as https://aka.ms/DowAVA, https://aka.ms/Lufthansa-zeroG, and https://aka.ms/TelstraAVA to solve problems such as chemical leak detection, optimizing aircraft turn-around, and traffic analytics. AVA has enabled DOW get closer than ever to reaching their safety goal of zero safety-related incidents. In the case of Lufthansa CityLine, AVA has enabled human coordinators to make better data-driven decisions to reduce aircraft turnaround time, thus reducing costs and improving customer satisfaction significantly. Telstra has been able to unlock new 5G business opportunities with the combination of Azure Video Analyzer, Azure Percept, and Azure Stack Edge.
AVA offers all capabilities that were available in LVA and more. A short summary of some of the prominent capabilities are as follows:
Process live video at the Edge
AVA Edge can be deployed on any Azure IoT Edge enabled X64 or AMD64 Linux device. Live video from existing cameras can be directed to AVA and processed on the edge device within the business environment’s network. In turn, this provides organizations with a solution that considers limited bandwidth and potential internet connectivity issues, all while maintaining the privacy of their environments.
Analyze video with AI of choice
As mentioned earlier, AVA can be integrated with Microsoft provided AI or with AI that is custom built to solve a niche problem, such as Intel’s https://docs.microsoft.com/azure/azure-video-analyzer/video-analyzer-docs/use-intel-grpc-video-analytics-serving-tutorial that provides capabilities such as detecting objects (a person, a vehicle, or a bike), object classification (vehicle attributions) and object tracking (person, vehicle and bike).
Azure AI is based on Microsoft’s https://www.microsoft.com/ai/responsible-ai-resources and the https://azure.microsoft.com/resources/transparency-note-live-video-analytics-lva/ provides additional guidance on designing response AI integrations.
Flexible video workflows
In addition to the https://docs.microsoft.com/en-us/azure/media-services/live-video-analytics-edge/media-graph-concept provided by LVA, the current release of AVA enables a variety of live video workflows via new building blocks such as https://docs.microsoft.com/azure/azure-video-analyzer/video-analyzer-docs/track-objects-live-video, https://docs.microsoft.com/azure/azure-video-analyzer/video-analyzer-docs/use-line-crossing, https://docs.microsoft.com/azure/azure-video-analyzer/video-analyzer-docs/pipeline#cognitive-services-extension-processor extension processor and https://docs.microsoft.com/azure/azure-video-analyzer/video-analyzer-docs/pipeline#video-sink
Simplify app development with easy-to-use widgets
AVA provides a https://github.com/Azure/video-analyzer/tree/main/widgets/readme.md which simplifies app development with a secure video player and enables visualization of AI metadata overlaid on video.
Azure Video Analyzer offers all these capabilities and more. Take a look at Azure Video Analyzer https://azure.microsoft.com/products/video-analyzer and https://docs.microsoft.com/azure/azure-video-analyzer/video-analyzer-docs/overview to learn more.
