ML Trends

Newsletter #2 - Mar 3, 2019

Where are the sharks?

We're back.

Since 1950s, 340,000 AI-related inventions have been patented. 49% of these patents relate to computer vision and the number is growing at a rate of 24% per year [source]. From face recognition to generating new faces, and saving lives through medical imaging to taking lives with autonomous weapons, computer vision is making a huge impact. Ready your armchair, get your coffee and let's talk about computer vision today.

Straight from the labs
Think tanks can kill.

The Army Contracting Command in the US has issued a call for industries and academia to submit proposals to help build its Advanced Targeting and Lethality Automated System, aka ATLAS. It will use AI and ML to give ground-combat vehicles, like tanks, autonomous targeting capabilities. ATLAS will use an algorithm to detect and identify targets and only parts of the fire control process will be automated. However, as required by law, a person will always be the one to actually make the decision to fire.

Ultimately, the driving force behind this idea is all about efficiency.

“Anytime you can shave off even fractions of a second, that’s valuable. A lot of engagement decisions in warfare are very compressed in time. If you’re in a tank and you see the enemy’s tank, they probably can also see you. And if you’re in range to hit them, they’re probably in range to hit you.”, said Paul Scharre, director of the Technology and National Security Program at the Center for a New American Society, a think tank based in Washington, D.C.

Here's the full article.

Into the realm of business
Excel's got superpowers.
For all the MS Excel geeks and fans out there, your life is about to get better. Microsoft announced a new feature - Insert Data from Picture. Point your smartphone camera at any document and convert it to an editable table within Excel. Imagine reading through financial or medical reports, invoices or orders and being able to instantly bring all that information to Excel - pretty cool! For now, it looks like the feature is going to work quite well on simple tables, but we really hope the algorithm is smart enough to deal with merged cells. That's one hard problem to solve.

Picture courtesy: [source]
Out in the real world
We found Waldo. Now let's find those sharks.
Shark detection plays an important role in ensuring the safety of water users. However, till date, this task has required manual intervention and depends on an appropriate vantage point, while the accuracy is affected by human fatigue and error.

Shark Spotters, a shark safety and research organisation based in Cape Town, and PatternLab, an information engineering service provider based in Switzerland, are teaming up to create an Automatic Shark Detection system. The system plans to use computer vision to detect sharks automatically. The data used to train the system will be sourced from previously collected footage as well as new footage collected from a decoy shark (see the image above). This application highlights the use of computer vision in a novel, non-invasive manner to ensure human safety.

Here's the full article.
Picture courtesy: Shark Spotters
We hope you enjoyed this edition of our newsletter. If you've got a couple of minutes to spare, do take this survey and help us get better.

If you found this interesting, you can help us grow by asking your friends, colleagues and enemies to subscribe to our newsletter.
Enjoy the rest of your weekend.

The ML Trends team
P.S. Recently, there's been a lot of discussion around taxes and AI, separately. Here's something that talks about both - Don't Fight the robots. Tax Them.
Copyright © 2019 ML Trends, All rights reserved.