Caffe
| Caffe | |
|---|---|
| Basic data
|
|
| Maintainer | Berkeley Artificial Intelligence Research (BAIR) |
| developer | Yangqing Jia |
| Publishing year | 2014 |
| Current version |
1.0 ( April 18, 2017 ) |
| operating system | Unixoide , macOS , Microsoft Windows |
| programming language | C ++ |
| category | Program library for deep learning |
| License | BSD license |
| German speaking | No |
| caffe.berkeleyvision.org/ | |
Caffe is a program library for deep learning . She was taught by Yangqing Jia during his Ph.D. -Time developed at the University of California, Berkeley's Vision and Learning Center .
Caffe has first the MATLAB - Implementation of fast Convolutional Neural Networks (CNN) for C and C ++ ported . Caffe contains numerous algorithms and deep learning architectures for the classification and cluster analysis of image data . CNN, R-CNN ( recurrent neural network ), LSTM ( long short-term memory ) and fully connected neural networks are supported. With Caffe, the graphics processor- based acceleration with cuDNN from Nvidia can be used, so that 60 million images can be processed per day.
As the primary programming are Python ( NumPy ) and MATLAB provided. Yahoo has integrated Caffe into Apache Spark ( caffeonspark ) in order to use deep learning in a distributed manner.
literature
- Yangqing Jia et al. a .: Caffe: Convolutional Architecture for Fast Feature Embedding . In: UC Berkeley EECS . Berkeley 2014 ( Online [PDF]).
Web links
Individual evidence
- ↑ Release 1.0 . April 18, 2017 (accessed April 23, 2018).
- ^ Embedded Vision Alliance: The Caffe Deep Learning Framework: An Interview with the Core Developers. In: Embedded Vision Alliance. Retrieved April 5, 2017 .
- ↑ Evan Shelhamer: Deep Learning for Computer Vision with Caffe and cuDNN. In: Nvidia. Nvidia Corp., October 15, 2014, accessed April 5, 2017 .
- ↑ Nikhil Ketkar: Deep Learning with Python . A hands-on introduction . Apress, 2017, ISBN 978-1-4842-2765-7 .
- ↑ Michael Thomas, Gabriela Motroc: CaffeOnSpark: Yahoo makes deep learning software open source. In: JAXenter. JAXenter, March 1, 2016, accessed April 5, 2017 .