Menoh

Menoh is DNN inference library with C API.

Menoh is released under MIT License.

Document

This codebase contains C API and C++ API.

Goal

DNN Inference with CPU
ONNX support
Easy to use.

Requirements

MKL-DNN Library (0.14 or later)
ProtocolBuffers

Build

Execute below commands in root directory.

sh retrieve_data.py
mkdir build && cd build
cmake ..
make

Installation

Execute below command in build directory created at Build section.

make install

Run VGG16 example

Execute below command in root directory.

./example/vgg16_example_in_cpp

Result is below

vgg16 example
-22.3708 -34.4082 -10.218 24.2962 -0.252342 -8.004 -27.0804 -23.0728 -7.05607 16.1343
top 5 categories are
8 0.96132 n01514859 hen
7 0.0369939 n01514668 cock
86 0.00122795 n01807496 partridge
82 0.000225824 n01797886 ruffed grouse, partridge, Bonasa umbellus
97 3.83677e-05 n01847000 drake

Please give --help option for details

./example/vgg16_example --help

Run test

Setup chainer

Then, execute below commands in root directory.

sh gen_test_data.py
cd build
cmake -DENABLE_TEST=ON ..
make
./test/menoh_test.out

Current supported operators

Activation functions

Elu
LeakyRelu
Relu
Softmax
Tanh

Array manipulations

Concat

Neural network connections

Conv
FC

Mathematical functions

Abs
Add
Sqrt

Normalization functions

BatchNormalization

Spatial pooling

AveragePool
GlobalAveragePool
GlobalMaxPool
MaxPool

License

Menoh is released under MIT License. Please see the LICENSE file for details.

Note: retrieve_data.py downloads data/VGG16.onnx. data/VGG16.onnx is generated by onnx-chainer from pre-trained model which is uploaded at http://www.robots.ox.ac.uk/%7Evgg/software/very_deep/caffe/VGG_ILSVRC_16_layers.caffemodel

That pre-trained model is released under Creative Commons Attribution License.

faithandbrave / menoh