OpenCV AI Kit aims to do for computer vision what Raspberry Pi did for hobbyist hardware – TechCrunch
A new gadget called OpenCV AI Kit (OAK) is designed to replicate the success of Raspberry Pi and other minimal computing solutions, but for the growing areas of computer vision and 3D perception. The new multi-camera circuit boards offer a lot of functions in a small open source unit and are now Looking for funds for Kickstarter.
The OAK devices use their cameras and the integrated AI chip to perform a number of computer vision tasks, e.g. For example, identifying objects, counting people, determining distances to and between objects in the image, and much more. This information is sent in a polished, ready-to-use form.
A reliable, low-cost, low-power computer vision device like this is a great blessing for anyone looking to build a smart device or robust robot that might otherwise require multiple and discrete cameras and other chips (not to mention some fumbling) with software).
Like the Raspberry Pi, which has become the first choice for hobby programmers who deal with hardware, pretty much everything about these devices is open source in the permitted MIT license. And it is officially associated with OpenCV, a widely used set of libraries and standards used in the computer vision world.
The actual device and the integrated AI were created by Luxonis, who previously created the CommuteGuardian, a kind of intelligent brake light for bicycles that tracks objects in real time to warn the rider. The team couldn’t find hardware that matched the bill, so they created their own and then worked with OpenCV to develop the successor to the OAK series.
There are actually two versions: the extra small OAK-1 and the triple camera OAK-D. They have many components in common, but the multiple camera units of the OAK-D mean that it provides a true stereoscopic 3D view instead of relying on other clues in the simple RGB image. These techniques are now better than ever, but real stereo is still a big plus. (The human vision system uses both in case you are wondering.)
The idea was to unify the computer vision system so that it does not have to be created or configured to get many projects off the ground faster. You can use built-in object and depth detection right away, or select and use the desired metadata to expand your own analysis of the 4K images (plus two 720p images) that are also received.
Very low power consumption also helps. Computer vision tasks can be quite demanding for processors and therefore use a lot of power, which is why a device like XNOR’s highly efficient chip was so promising (and why this company was picked up by Apple). The OAK devices do not take XNOR to the extreme, but with a maximum power consumption of a handful of watts, they can be operated with normal-sized batteries for days or weeks, depending on the task.
The details will undoubtedly be interesting for those who know the pros and cons of such things – ports and cables, as well as GitHub repositories, etc. – but I won’t duplicate them here as they are all copied in the campaign in an orderly fashion . Here is the quick version:
If this seems to be something that your project or laboratory could use, you may want it Get on the Kickstarter quickly, as there are some early discounts for early risers and the retail price will double. $ 79 for the OAK-1 and $ 129 for the OAK-D sound like bargains to me, based on their skills (they’ll eventually be $ 199 and $ 299, respectively). And Luxonis and OpenCV are hardly nocturnal organizations that deal with vaporware, so you can support the campaign with confidence. They also flew past their destination in an hour, so you didn’t have to worry about it.