Computer Vision Library: the only type of library without books
In this articles, we explain you the meaning of a computer vision library and we give you some examples, for illustrative purpose.
So, what is a Computer vision Library?
A library refers to a set of mathematical functions that can be directly used in a computer program. In computer vision, you can find libraries that are made to build neural network for machine learning.
The libraries are created mainly for developers. If you are looking for quality libraries, you should look into the different frameworks available online.
To make it simple, a framework is a toolbox to train neural networks. It includes libraries and different types of programs. Moreover, they are not only used by developers but also by data scientists.
Examples of Framework with the best libraries
We made, for you, a shortlist of the frameworks we consider to be the best and that can offer you satisfying computer vision libraries:
- Caffe: originally developed at UC Berkeley, it is now largely used in the research community. It has a focus on image processing and object recognition. Facebook released Caffe2 in April 2017, which is a more lightweight, modular and scalable version of Caffe.
- Tensorflow: created by the GoogleBrain team, it was released in November 2015. Its capabilities are quite generic and it is heavily supported by the Google community. It has known a strong growth and it is now by far the most popular framework used by deep learning scientists to experiment and put neural networks in production.
- Deeplearning4j: is an independant deep learning programming library written for Java and the Java virtual machine. It is also a computing framework with wide support for deep learning algorithms.
- CNTK (now called Microsoft Cognitive Toolkit) and MXNet: developed by Microsoft and Amazon respectively, these two frameworks are fully integrated with their respective clouds.
- Torch: used by 2 of the Big Four, Google and Facebook, it is a framework that offers support for machine learning algorithms. It is an easy-to-use toolbox, based on LuaJIT. If you heard about Torch, you might have heard about Pytorch, which relies on Torch library. It is mainly used in deep learning and runs on Python, as opposed to Torch.