viernes, 24 de febrero de 2012

Sopla más rápido, maldita

Seguro que has visto alguna película en la que el malo pone a alguien colgando de una cuerda a punto de caer en una piscina llena de tiburones/pirañas/cocodrilos/señoras de las rebajas. Y entonces pone una vela quemando lentamente la cuerda y se va.

Pues yo tengo esa misma sensación, pero en vez de una piscina debajo hay un balate y en vez de tener que escaparme me han dado una colchoneta hinchable para amortiguar la caída. Así que empiezo a soplar y a soplar. Pero mi capacidad pulmonar no da pa mucho (el asma, ya sabes). Mientras, la vela sigue quemando la cuerda incansable. Y por mucho que sople no consigo que la colchoneta se hinche tan rápido como debería.

Me paso el día diciéndome "Sopla más rápido, maldita, que no llegas"... Ya veremos si llego o no... Pero id preparando el cepillo y el recogedor, porque me veo cayendo sin colchoneta debajo.

viernes, 17 de febrero de 2012

PhD... for dummies

When you write your PhD dissertation you are supposed to be an expert in your field, so you write it showing your knowledge in that field. That's one of the reasons you have a chapter introducing the tools you've used (algorithms and models, mainly).

What if you are not such an expert? I'm an engineer. Tell me a problem and I'll solve it. I'll look for algorithms doing something similar, compare them and modify them to solve the problem. If there's no algorithm available to solve your problem, I'll combine some of them, innovating a solution. Moreover, I can analyse the results, measure execution time, make some fancy graphs and even write the process down for somebody else to reproduce it (and in LaTeX, ladies and gentlemen).

But I'm not a scientist. I don't really care why, mathematically talking, one algorithm is better than another. I don't mind why the second derivative of the Gaussian filter is equivalent to convolve the image with a given mask. I could do some maths that I don't fully understand and extract one formula from another until I can prove that my solution is correct... or even copy them from Wikipedia... but I don't really care.

I'm an engineer. I analyse a problem, extract requirements and solve them. I'm writing my "Introduction to Computer Vision" for dummies: algorithm names, when to use them and how to implement them. And I'm proud of it.