In the course of these months/years since the last article on this blog, many things have happened, first of all the advent of Covid with all the consequences it has brought about restrictions and worries and secondly job changes but I won’t stay here to dwell on these facts. I just want to say that there have been some difficulties which however have left room and time for the development of software projects. I have continued to maintain and develop ARnft and jsartoolkitNFT related to our open source organization WebARKit, and in parallel carry out other research.
std::vector to interpret the underlying data. Also in this case not all the classes have been implemented nor all the functions. Furthermore, exhaustive examples and documentation are still missing. Time permitting, I hope to fill this gap and also present satisfactory tests.
At the same time since in the case of webarkit-jsfeat-cpp it meant rewriting a project from scratch, i looked at other libraries that could meet the needs of the project as a base or element to be integrated. Among many that i probed, including OpenCV it seemed convenient dlib you can find the documentation here dlib.net. Dlib offers over and over algorithms useful for image processing and computer vision also algorithms for Machine Learning, but this will not be the reason to choose it. From a first taste I could see that it is very simple to apply transformations to images. You can see it in my repository dlib_em.js The difference I saw is that the final library in wasm is much smaller if it was compiled with OpenCV. An interesting feature is that dlib contains several optimizations for matrix computation that are difficult to develop from scratch. So I think he’s a very attractive candidate. I think I will start a prototype based on dlib and then compare it to webarkit-jsfeat-cpp. This will be my goal for the next few months, time permitting of course. We will certainly see some good ones. See you soon and happy coding!