SPOTIFY CODE CODE
Well, I've not classified code rates like that, but it would depend on the BER (bit error rate) of the communications channel, how many errors your communication can tolerate, and the type of FEC you're doing. The model is quite simple, only a few hundred parameters, so I perform the neural network layer multiplications myself using just some for loops / tanh. I extract a bunch of semi-arbitrary features from the image (width/height ratio of logo, relative distance to the first and last dot, center line size, etc) and I trained a model to produce deskewing parameters from that. And then it worked quite well so it seems like people forgot it was only an undocumented feature =) There's actually a small neural network ML model in the scanner as well. But then as a test, to see if I could improve the scanning, I added a logo based detector as well that first finds the logo then finds the waveform. Originally we required the code to have a distinct rectangular border around (to make deskewing simpler, for when the camera is angled). Obviously encoding a whole Spotify URI/URL would require way more bits meaning that the code would look more intrusive. We had to strike a balance between making the code look nice, and Spotify thematic, while at the same time fitting enough information to encode any Spotify resource. Some more info, I made the scannable codes back in 2016, and haven't really worked on it since, so the article brings back memories. Are you interested in promoting your own content? STOP! Read this first.
That means no image posts, no memes, no politics.Please keep submissions on topic and of high quality.r/programming is a reddit for discussion and news about computer programming
0 kommentar(er)
