This week in AI & Machine Learning: a how-to for Classification Models, the ethics of emotion-detection AI, a better way to capture lightning and more!

A Note from the Author 

I’m Paul Davenport, Plainsight’s Director of Communications! I’ll be dishing out your roundup of AI & ML headlines again this week while your regular AI expert Bennett enjoys his well-deserved vacation.

Did you know that 92.5 percent of photos are captured using smartphones? Or that more than 90 percent of the population has one of these camera-equipped devices at the ready? 

In our latest blog, we show you how to leverage Classification Models for better organizing any collection of images—whether you’re a professional archivist or just a hiker with a keen eye—using tools and techniques that create highly accurate models with relative ease. 

Artificial Intelligence News

The pros and cons of emotion-detection AI

While many of the more practical, solution-based applications for computer vision (ie. mask detection, supply chain monitoring) are designed to specifically avoid ethical dilemmas, both Microsoft and Google have continued to explore AI-based emotion recognition technology that many experts worry could be overly invasive—if not outright dangerous. 

Microsoft recently restricted the general use of an AI-driven cloud software tool that the company claims can be used to infer people’s emotions. Despite these general-use restrictions, Microsoft will continue leveraging this technology in the Seeing AI app, a product designed to help visually impaired users communicate with more nuance. Google, on the other hand, still offers software that claims to detect human emotions as part of its ML Kit tool, though Google spokespeople have walked back the capabilities recently as other companies faced backlash for using similar technologies. 

How Synthetic Data is making vision AI even smarter

Ever wonder how self-driving cars got so wise to the ways of the road years before they were legally allowed to hit the streets? What about the ways computer vision solutions for preventive maintenance learned to predict potential risks and flaws in production processes? Thanks to the use of synthetic data, these and myriad other vision AI scenarios have taken flight. A recent article from Forrester digs into the existing and future ways in which synthetic data can help enable ever-smarter AI. 

Vision AI that’s looking toward the sky

Capturing lightning in a bottle may still be out of humanity’s grasp, but we’re closer than ever to capturing clear imagery of lightning strikes thanks to a new open-source computer vision initiative. Called ZapCapture, the project runs on a tiny netbook (for easy field deployment) and features Python 3 capabilities to help storm chasers and photographers spend less time manually sorting through their footage. While the tool is unlikely to shed light on how exactly lightning is created—a mechanism that’s still not widely understood today—it will undoubtedly result in more striking imagery of this eye-catching natural phenomena.  

Join our Community

See you next week! Until then, keep the conversation going on Plainsight’s AI Slack Channel


About the Author & Plainsight

Paul Davenport hails from Boston and has an extensive background in tech journalism. He helps Plainsight in its mission to make vision AI accessible to entire enterprise teams.

Plainsight’s vision AI platform streamlines and optimizes the full computer vision lifecycle. From data annotation through deployment, customers can quickly create and successfully operationalize their own vision AI applications to solve highly diverse business challenges.

 View All Blogs