computer vision

i'm starting my journey into computer vision. where should i start, what books/documentation should i check out? is there a library that's open and good?

>>3151 OpenCV is a great BSD licensed library for starting, there is; however, a caveat. Performance is a order magnitude better on Intel processors than non-Intel processors. Which is a deal breaker for me, but hey if you want to jump in and play OpenCV is great. As for books, I cannot recommend any as I have not read any, though machine learning is a driving factor in this area so ones tied into that and computer vision would probably be he optimal ones.

Definitely OpenCV http://www.bogotobogo.com/cplusplus/files/OReilly%20Learning%20OpenCV.pdf Open Source, well maintained, wrappers for a bunch of languages, lots of features etc ...

>>3156 There is a new version of the book for OpenCV 3 though I haven't looked at it

>>3151 Opencv is pretty much the standard, the alternative is matlab. Reference for computer vision theory. http://www.cse.psu.edu/~rtc12/CSE486/ Guides on setting up your first python opencv applications on raspberry pi. http://www.pyimagesearch.com/ This is going to sound pretty corny, but after a few years of learning computer vision it felt like studying philosophy more than computer science.

I'm quite new to computer vision myself but for the classical approach this books seems fine: http://szeliski.org/Book/ However, It seems like most of the modern stuff is done with Convolutional Neural Networks. I remember someone called it the new "master algorithm" for computer vision. This semester I heard some talks form this field and always such CNNs were involved. Also, my professor said that the field progresses really fast and there is currently no comprehensive book for this modern approach to computer vision. In my opinion that's a quite nice development because such CNNs are relatively easy to understand and also easy to use with the current frameworks. If I find the time I look at the popular papers from this field and often they are easier to understand than the classical approach with all the complicated modelling and optimization. (But I'm also working at understanding this better.)

Ive worked a lot with Computer Vision. As for programming languages: If you are a beginner, then I recommend python with opencv As for books, for the start I would recommend you: >http://cvrs.whu.edu.cn/downloads/ebooks/Multiple%20View%20Geometry%20in%20Computer%20Vision%20(Second%20Edition).pdf This book mainly deals with the geometry and math. It is very dry to read. The knowledge is very theoretical, and hence often not directly necessary for easier practical applications, but still, if you want to master the field than you will have to learn these things at one point. >https://www.amazon.de/Introductory-Techniques-3-D-Computer-Vision/dp/0132611082 This is the first book that taught me computer vision. Its really good, but unfortunately by now it is also very old. Therefore, it is missing a lot of newer things. Still I would recommend it. >A book about neural networks, for example the one from Ian Goodfellow A lot of modern research in computer vision is done via neural networks, hence you should learn it. There are not that many good books about this field. The best one in my opinion is from Ian Goodfellow, but it is more about general neural networks, and less about their application to computer vision. >https://www.amazon.de/Learning-OpenCV-Computer-Vision-Library/dp/0596516134 This might be interesting, but I never read it, so I can not judge its quality.