As for His Computer Equipment, It Was Confiscated

msum

Hello ! I just read this afternoon what TubeTagger was And know that this video recognition software learns to recognize images by analyzing clips on YouTube and the tags associated with them! In my opinion, it's quite stunning! The description : The program developed by the team from the University of Kaiserslautern in Germany modifies the “training” usually carried out by this type of software: products dedicated to video recognitionusually optimize themselves by browsing a library of movie samples. What until now meant to title each of them by hand.

TubeTagger meanwhile learns to add keywords to videos image manipulation service by watching YouTube. The operation is as follows: the researchers give the keyword “football” to TubeTagger. The latter then automatically downloads fifty videos on YouTube that Internet users have already labeled with this tag, and it examines the colors and movements of each of these contents. In my opinion, this kind of automated process (where we do without humans to carry out long BUT complex tasks) is part of the logical development of the Internet. Indeed, the staggering amount of data that users deposit on servers then serve as databases.


It “suffices” then to design algorithms synthesizing the information available on the servers to be able, as in this case, to identify a video without the human eye even looking at it. Just by comparison. In short, let's continue to feed our internet databases for better. and for worse! In short, I find it silly, even stupid , simply for several reasons. Ok, we don't modify Wikipedia to talk nonsense, it's not good, that's why Wikipedia hunts down malicious editors. Laborious work that actually reduces the number of videos used for this necessary optimization.