| 1 |
Source: keras-applications |
| 2 |
Maintainer: Debian Science Maintainers <[email protected]> |
| 3 |
Uploaders: |
| 4 |
Stephen Sinclair <[email protected]>, |
| 5 |
Section: science |
| 6 |
Build-Depends: |
| 7 |
debhelper-compat (= 13), |
| 8 |
dh-sequence-python3, |
| 9 |
python3-all, |
| 10 |
python3-numpy, |
| 11 |
python3-setuptools, |
| 12 |
Standards-Version: 4.7.3 |
| 13 |
Vcs-Browser: https://salsa.debian.org/science-team/keras-applications |
| 14 |
Vcs-Git: https://salsa.debian.org/science-team/keras-applications.git |
| 15 |
Homepage: https://keras.io/ |
| 16 |
|
| 17 |
Package: python3-keras-applications |
| 18 |
Architecture: all |
| 19 |
Section: python |
| 20 |
Depends: |
| 21 |
${misc:Depends}, |
| 22 |
${python3:Depends}, |
| 23 |
Recommends: |
| 24 |
python3-keras, |
| 25 |
Description: popular models and pre-trained weights for the Keras deep learning framework |
| 26 |
Keras is a Python library for machine learning based on deep (multi- |
| 27 |
layered) artificial neural networks (DNN), which follows a minimalistic |
| 28 |
and modular design with a focus on fast experimentation. |
| 29 |
. |
| 30 |
Features of DNNs like neural layers, cost functions, optimizers, |
| 31 |
initialization schemes, activation functions and regularization schemes |
| 32 |
are available in Keras a standalone modules which can be plugged together |
| 33 |
as wanted to create sequence models or more complex architectures. |
| 34 |
Keras supports convolutions neural networks (CNN, used for image |
| 35 |
recognition resp. classification) and recurrent neural networks (RNN, |
| 36 |
suitable for sequence analysis like in natural language processing). |
| 37 |
. |
| 38 |
It runs as an abstraction layer on the top of Theano (math expression |
| 39 |
compiler) by default, which makes it possible to accelerate the computations |
| 40 |
by using (GP)GPU devices. Alternatively, Keras could run on Google's |
| 41 |
TensorFlow (not yet available in Debian). |
| 42 |
. |
| 43 |
Keras Applications is the applications module of the Keras deep |
| 44 |
learning library. It provides model definitions and pre-trained |
| 45 |
weights for a number of popular architectures, such as VGG16, ResNet50, |
| 46 |
Xception, MobileNet, and more. |