From real time crime detection using video surveillance to audio induced predictive maintenance, organizations are leveraging Artificial Intelligence to drive critical use cases. The concept of digital and business transformation has been around for a couple years but now we are seeing businesses embrace AI to propel their strategy forward. As new use cases are added to their AI agenda, vendors need to add new tools and improvements to their platforms to keep up with customer needs. Microsoft is one of those companies that has been adding improvements to their Azure AI that focuses on making it easy for data scientists and developers to deploy, manage and secure AI workloads.
This morning, Microsoft released some pretty cool new features and capabilities to their Azure Cognitive Services. The new features they announced show their commitment to providing their customers with the latest enhancements and tools that allow for the development of cutting edge AI use cases that organizations are looking for. The first thing announced was preview of a new Anomaly Detector Service, which is an AI based service that allows for companies to identify problems quicker and with a better degree of accuracy. Their other announcement and probably my favorite is the general availability for their Custom Vision that allows for the better detection and identification of objects in images. Image recognition and classification has to be one of the hottest applications of AI right now and Custom Vision provides a much easier way for developers to build and deploy image classifiers.
What is Cognitive Services
For those of you who are unfamiliar with Cognitive Services, they are Microsoft’s custom built AI capabilities and Machine Learning services, API’s, and SDK’s that allow the customer to build and deploy these services without having direct ML / AI skills.There are five main areas of focus for Cognitive Services: Vision, Speech, Search, Language, and Knowledge. Cognitive Services make it easier for developers to add in AI services into their applications by adding in. The goal of Cognitive Services is to allow developers to build applications that can see, hear, speak, and reason.
Each of the five pillars of the Cognitive Services break have multiple API’s that developers can use to achieve various tasks by simply making the API Call. Here are a few examples of each of the pillars:
- Vision – Computer Vision, Content Moderation, Facial Recognition
- Speech – Translator Speech, Bing Search Speech, Speaker Recognition
- Search – Allows you to search through Bing using various tags and titles
- Language – Text Analytics, Spell Check, Translator
- Knowledge – You can use your semi structured data to build a QnA engine.
In the world of democratizing AI, Microsoft claims that more than 1 Million developers have already discovered and tried Cognitive Services to add intelligent capabilities to their applications.
Lets dive into their announcements a little further
Anomaly Detection as an AI service
This is a new service being offered in the Cognitive Services family and its goal is to provide organizations with the ability to identify unusual patterns or rare events within your data that might equate to negative impacts to the business. Fraud is the first use case that comes to mind but detecting unusual patterns and outliers in the data can help any business in any vertical from Oil & Gas to manufacturing. With Anomaly Detection, through a single API, developers can build in detection capabilities into their apps. I think this is a great service that Microsoft is now offering to their users. I know Microsoft has been using anomaly detection internally for a while and by packing that for customers, they extend capabilities that could greatly impact a business.
Custom Vision
Powered by machine learning, Custom Vision (now generally available) allows for organizations and developers to build and train their own image classifiers. I think the most impressive thing about Custom Vision is that it accelerates the time it takes for organizations to build, train, deploy and improve custom image classifiers. Without tools like this, most organizations would not have the skill sets to successful build one of these but through this service, a broader subset of users can contribute to building classifiers. Once you build your classifier, you can explore them, run them off-line or in real time depending on your use case. With the GA announcement today, Custom Vision is delivering the following improvements around Higher Quality, Simplicity and Train in the Cloud, run anywhere.

Want to learn more about custom vision? Check out their product page
To learn more and start building vision and search intelligent apps, please visit the Cognitive Services site to get started.
I am pretty excited about these new announcements and can’t wait to start playing with Custom Vision. Remember Cory’s AI Poo Patrol blog? This could be the perfect tool for him to finally finish his AI project. I am also very bullish on the new additions and features that Microsoft is adding to their AI platforms. This represents Microsoft’s continued commitment to providing developers and data scientists with the latest tools and services for them to be successful. Kudos Microsoft, we can’t wait to see what is next!
Your bearded friends Brett & Kyle
Be the first to write a comment.