This is one of the many experiments I'm doing as part of my Master's thesis in Computer Engineering.
I'm working on an unsupervised bayesian motion detector based on features clustering and coherent motion likelihood, as suggested by Brostow/Cipolla. The features I'm tracking here are the Rosten/Drummond corners. For references, just ask me.
This video is the output of the software I'm writing in C/C++. I also wrote the code for features detection and optical flow, although based on the sources provided by Rosten and Drummond for the first, and on OpenCV for the second.
Oh, the software will be released open-source through the GPL license as soon as it's more stable.
fa parte della famosa roba che ci servirà in futuro per caso?
shurzan 4 years ago
certo, ma tu non sai ancora che ideazze che ho.. le classiche idee stupide ma funzionali e fiche... 俺は鬼才
iron3 4 years ago