OpenCV 5 Is Here: The Biggest Leap in Years for Computer Vision

Post by Joel_Mckay » Wed, Jun 10, 2026, 5:56 AM UTC

akssri wrote: post #135

Joel_Mckay wrote: post #70

Earlier quoted context omitted.

Done a few projects with OpenCV over the years, and I agree it can be fun. However, it has a few issues: 1. Patented algorithms that are effectively impossible to license in a commercial setting. 2. Permuted API that change how identically named functions behave over versions. 3. Hardware CUDA version coupling deprecating support every major release. 4. Inconsistent and contradictory documentation in the constant sub…

Yup, it's basically the ROS of computer-vision.

Post by Joel_Mckay » Wed, Jun 10, 2026, 6:33 AM UTC

jrk wrote: post #132

Joel_Mckay wrote: post #101

Earlier quoted context omitted.

Machine vision has always been resource intensive... and if you are doing trained ML projects the hardware choices are actually very limited. To enable Intel TBB, CUDA, and CPU specific compiler optimizations... one will almost certainly need to re-build the library, and customize your application build. Some tasks degrade in performance on a GPU, and others are 740 times faster... ymmv. =3

It's not that you need to turn on some extra library backends and rebuild, it's that the abstractions themselves are fundamentally at odds with hitting peak performance on many things so you have to rewrite your code. Individual image processing operations are often very low arithmetic intensity. If you don't combine them into much larger subroutines—which are necessarily less generic and orthogonal—you spend all you…

Post by Joel_Mckay » Wed, Jun 10, 2026, 9:14 AM UTC

zdkl wrote: post #144

Joel_Mckay wrote: post #119

Earlier quoted context omitted.

There is a CLI photogrammetry OSS project with rather litigious faculty members behind the code. However, at least that group was upfront about what was expected of the library users, and didn't do something dodgy like quietly merge it into another community library like OpenCV. I discovered that while porting it to a Pi ARM platform years ago (yes it was slow... lol.) Forgot when the IP becomes public domain, but yo…

They wouldn't happen to be french cartography would they?

Post by zipy124 » Wed, Jun 10, 2026, 9:52 AM UTC

fzysingularity wrote: post #140

plasticeagle wrote: post #37

The thing I love about OpenCV is that it remains hands down the best library for simply loading images and video. I've never even used any of its fancy computer vision features, but if I need to load a video file and look at the pixels - which I did need to do recently for an art project - OpenCV does it in about four lines of code.

That’s a pretty large binary for simply loading images. In all honesty, opencv has stood the test of time and I’m certain newer LLMs will likely not attempt to rewrite it from scratch. P.S. I’ve been a user since the IplImage days, circa 2007, and I’d still consider using it over most CV libraries today.

Post by killingtime74 » Wed, Jun 10, 2026, 10:46 AM UTC

GreenSalem wrote: post #51

AI written release post and it shows...