Do you believe Video Analytics is an expensive science fiction technology that is only used for research? A technology that will some day fall from the sky and solve many of the problems we have today? (Did someone say Minority Report?)

Well, I totally understand you, but luckily that is not the case. First, let’s ground ourselves on what my definition of Video Analytics is. If you boil it down, Video Analytics is really “just" using computer vision - which is a field of study that teaches computers how to “see” - techniques across video frames to extract value from digital video content.

OK, maybe I am oversimplifying it, but good solutions that meet your needs will work and look “simple.” The most advanced solutions include multiple deep learning models implemented into an algorithm pipeline (think Machine Learning on steroids) to solve complex solution needs.

Based on most of the cases I have solved with Video Analytics, here is a definition, expressed as an equation: Video Analytics = computer vision + machine learning + real time analysis.

If you are new to Video Analytics, you may be focusing on all of the cool things it can do about analyzing a person or object. But the true value is the ability to capture new data dimensions you can use with existing data to build on your understanding of a given problem statement.

Think of it this way: You may have some industrial product manufacturing process that has hundreds of sensors in place that monitor everything in sub second intervals. Yet still, you have high rates of product failure. Even with all that data you cannot identify causality. But when you add Video Analytics to monitor the product as it is being produced, you can observe the production defect as it happens and tie it to your existing sensor data to easily identify root cause.

You would be surprised to see how much university research that is now being commercialized in this space. There are models that can identify and track thousands of everyday objects, determine if workers are wearing required safety gear or even read your micro expressions to understand your customer satisfaction of a delivered service. And much, much more…

I have lead projects that incorporate lasers and different light spectrums into industrial and retail applications, captured age and gender demographics at a festival and tied it to geolocation data, implemented real time marketing content driven by facial analytics and conducted sentiment and people flow analysis for a world leading public transportation hub.

This is just the beginning. There is a huge amount of ongoing research in data science around model architectures and model development workflows, and the hardware vendors are applying the same energy to their products to try and keep up with these demands.

And it has been building for decades. Think about this: If it takes 20 years for a cutting-edge research technology to get to market, we are going to see a tsunami of Video Analytic solutions hitting the market in the next 10 years.

The first wave of these solutions have already made land. Just have a look at Amazon, Microsoft, Google or IBM as examples that offer services you can integrate with to analyze images. And there are many more niche players out there offering the same and, in some cases, more.

When you take components of the extensive research output available and apply them, you can do amazing things. For example, you can detect patterns that are invisible to the naked eye to improve quality and maintenance activities or monitor a patient to capture new data on pain tolerances and emotional states that could improve treatment.

The only question you are left with is not whether Video Analytics is science fiction or reality. The question is: How and where can Video Analytics create value for you?

Key takeaways

Video Analytics is already here. It can do much more and solve many more use cases than you can imagine. There are both solutions and integration services available today that can enable your business.

About this author

Picture of Edward Mauser

Edward Mauser

Head of Technology – Video Analytics

Edward Mauser has worked with product and service development with emerging technologies for two decades. His passion for technology and thirst for new knowledge has made him a student for life.

Tilføj kommentar

Comment editor

  • No HTML tags allowed.
  • Lines and paragraphs break automatically.
Regler og vilkår for brug af bloggen