## Refer to http://caffe.berkeleyvision.org/installation.html

# Contributions simplifying and improving our build system are welcome!

# cuDNN acceleration switch (uncomment to build with cuDNN).

# USE_CUDNN := 1

# CPU-only switch (uncomment to build without GPU support).

CPU_ONLY := 1

# Enable ACL (ARM Compute Library)

USE_ACL :=1

USE_PROFILING := 0

ifeq ($(USE_ACL), 1)

ifeq ($(ACL_ROOT),)

$(error ACL_ROOT does not specified. use "export ACL_ROOT='path of acl soure code'")

endif

ACL_INCS :=$(ACL_ROOT)/include

ACL_INCS +=$(ACL_ROOT)

ACL_LIBS_DIR :=$(ACL_ROOT)/build

ACL_LIBS_DIR +=$(ACL_ROOT)/build/arm_compute

ACL_LIBS :=arm_compute OpenCL

endif

# uncomment to disable IO dependencies and corresponding data layers

# USE_OPENCV := 0

# USE_LEVELDB := 0

# USE_LMDB := 0

# uncomment to allow MDB_NOLOCK when reading LMDB files (only if necessary)

# You should not set this flag if you will be reading LMDBs with any

# possibility of simultaneous read and write

# ALLOW_LMDB_NOLOCK := 1

# Uncomment if you're using OpenCV 3

# OPENCV_VERSION := 3

# To customize your choice of compiler, uncomment and set the following.

# N.B. the default for Linux is g++ and the default for OSX is clang++

# CUSTOM_CXX := g++

#CUSTOM_CXX := aarch64-linux-gnu-g++

#os :=linux

#arch :=arm64-v8a

# CUDA directory contains bin/ and lib/ directories that we need.

CUDA_DIR := /usr/local/cuda

# On Ubuntu 14.04, if cuda tools are installed via

# "sudo apt-get install nvidia-cuda-toolkit" then use this instead:

# CUDA_DIR := /usr

# CUDA architecture setting: going with all of them.

# For CUDA < 6.0, comment the *_50 through *_61 lines for compatibility.

# For CUDA < 8.0, comment the *_60 and *_61 lines for compatibility.

CUDA_ARCH := -gencode arch=compute_20,code=sm_20 \

-gencode arch=compute_20,code=sm_21 \

-gencode arch=compute_30,code=sm_30 \

-gencode arch=compute_35,code=sm_35 \

-gencode arch=compute_50,code=sm_50 \

-gencode arch=compute_52,code=sm_52 \

-gencode arch=compute_60,code=sm_60 \

-gencode arch=compute_61,code=sm_61 \

-gencode arch=compute_61,code=compute_61

# BLAS choice:

# atlas for ATLAS (default)

# mkl for MKL

# open for OpenBlas

#BLAS := atlas

BLAS := open

# Custom (MKL/ATLAS/OpenBLAS) include and lib directories.

# Leave commented to accept the defaults for your choice of BLAS

# (which should work)!

# BLAS_INCLUDE := /path/to/your/blas

# BLAS_LIB := /path/to/your/blas

# Homebrew puts openblas in a directory that is not on the standard search path

# BLAS_INCLUDE := $(shell brew --prefix openblas)/include

# BLAS_LIB := $(shell brew --prefix openblas)/lib

# This is required only if you will compile the matlab interface.

# MATLAB directory should contain the mex binary in /bin.

# MATLAB_DIR := /usr/local

# MATLAB_DIR := /Applications/MATLAB_R2012b.app

# NOTE: this is required only if you will compile the python interface.

# We need to be able to find Python.h and numpy/arrayobject.h.

PYTHON_INCLUDE := /usr/include/python2.7 \

/usr/lib/python2.7/dist-packages/numpy/core/include

# Anaconda Python distribution is quite popular. Include path:

# Verify anaconda location, sometimes it's in root.

# ANACONDA_HOME := $(HOME)/anaconda

# PYTHON_INCLUDE := $(ANACONDA_HOME)/include \

# $(ANACONDA_HOME)/include/python2.7 \

# $(ANACONDA_HOME)/lib/python2.7/site-packages/numpy/core/include

# Uncomment to use Python 3 (default is Python 2)

# PYTHON_LIBRARIES := boost_python3 python3.5m

# PYTHON_INCLUDE := /usr/include/python3.5m \

# /usr/lib/python3.5/dist-packages/numpy/core/include

# We need to be able to find libpythonX.X.so or .dylib.

PYTHON_LIB := /usr/lib

# PYTHON_LIB := $(ANACONDA_HOME)/lib

# Homebrew installs numpy in a non standard path (keg only)

# PYTHON_INCLUDE += $(dir $(shell python -c 'import numpy.core; print(numpy.core.__file__)'))/include

# PYTHON_LIB += $(shell brew --prefix numpy)/lib

# Uncomment to support layers written in Python (will link against Python libs)

# WITH_PYTHON_LAYER := 1

# Whatever else you find you need goes here.

INCLUDE_DIRS := $(PYTHON_INCLUDE) /usr/local/include

LIBRARY_DIRS := $(PYTHON_LIB) /usr/local/lib /usr/lib

# If Homebrew is installed at a non standard location (for example your home directory) and you use it for general dependencies

# INCLUDE_DIRS += $(shell brew --prefix)/include

# LIBRARY_DIRS += $(shell brew --prefix)/lib

# NCCL acceleration switch (uncomment to build with NCCL)

# https://github.com/NVIDIA/nccl (last tested version: v1.2.3-1+cuda8.0)

# USE_NCCL := 1

# Uncomment to use `pkg-config` to specify OpenCV library paths.

# (Usually not necessary -- OpenCV libraries are normally installed in one of the above $LIBRARY_DIRS.)

# USE_PKG_CONFIG := 1

# N.B. both build and distribute dirs are cleared on `make clean`

BUILD_DIR := build

DISTRIBUTE_DIR := distribute

#HDF5

USE_HDF5 := 1

HDF5_INCLUDE_DIRS := /usr/include/hdf5/serial

HDF5_LIBRARY_DIRS := /usr/lib/aarch64-linux-gnu/hdf5/serial

HDF5_LIBRARIES :=hdf5_hl hdf5

# Uncomment for debugging. Does not work on OSX due to https://github.com/BVLC/caffe/issues/171

# DEBUG := 1

# The ID of the GPU that 'make runtest' will use to run unit tests.

TEST_GPUID := 0

# enable pretty build (comment to see full commands)

Q ?= @


[](LICENSE)

Please refer to [CaffeOnACL Release NOTE](https://github.com/OAID/caffeOnACL/blob/master/acl_openailab/Reversion.md) for details

sudo apt-get -y update

sodo apt-get -y upgrade

sudo apt-get install -y build-essential pkg-config automake autoconf protobuf-compiler cmake cmake-gui

sudo apt-get install -y libprotobuf-dev libleveldb-dev libsnappy-dev libhdf5-serial-dev

sudo apt-get install -y libatlas-base-dev libgflags-dev libgoogle-glog-dev liblmdb-dev libopenblas-dev

sudo apt-get install -y libopencv-dev python-dev

sudo apt-get install -y python-numpy python-scipy python-yaml python-six python-pip

sudo apt-get install -y scons git

sudo apt-get install -y --no-install-recommends libboost-all-dev

pip install --upgrade pip

Recommend creating a new directory in your work directory to execute the following steps. For example, you can create a direcotry named "oaid" in your home directory by the following commands.<br>

cd ~

mkdir oaid

cd oaid

git clone https://github.com/ARM-software/ComputeLibrary.git

git clone https://github.com/OAID/caffeOnACL.git

git clone https://github.com/google/googletest.git

cd ~/oaid/ComputeLibrary

scons Werror=1 -j8 debug=0 asserts=1 neon=1 opencl=1 embed_kernels=1 os=linux arch=arm64-v8a

export ACL_ROOT=~/oaid/ComputeLibrary

cd ~/oaid/caffeOnACL

cp acl_openailab/Makefile.config.acl Makefile.config

make all distribute

cd ~/oaid/googletest

cmake CMakeLists.txt

make

sudo make install

export CAFFE_ROOT=~/oaid/caffeOnACL

cd ~/oaid/caffeOnACL/unit_tests

make clean

make

If the output message of the following two tests is same as the examples, it means the porting is success.

export LD_LIBRARY_PATH=~/oaid/caffeOnACL/distribute/lib:~/oaid/ComputeLibrary/build

cd ~/oaid/caffeOnACL/data/ilsvrc12

sudo chmod +x get_ilsvrc_aux.sh

./get_ilsvrc_aux.sh

cd ../..

./scripts/download_model_binary.py ./models/bvlc_reference_caffenet

./distribute/bin/classification.bin models/bvlc_reference_caffenet/deploy.prototxt models/bvlc_reference_caffenet/bvlc_reference_caffenet.caffemodel data/ilsvrc12/imagenet_mean.binaryproto data/ilsvrc12/synset_words.txt examples/images/cat.jpg

output message --

---------- Prediction for examples/images/cat.jpg ----------

0.3094 - "n02124075 Egyptian cat"

0.1761 - "n02123159 tiger cat"

0.1221 - "n02123045 tabby, tabby cat"

0.1132 - "n02119022 red fox, Vulpes vulpes"

0.0421 - "n02085620 Chihuahua"

cd ~/oaid/caffeOnACL/unit_tests

./test_caffe_main

output message:

[==========] 29 tests from 6 test cases ran. (1236 ms total) [ PASSED ] 29 tests.

[](LICENSE)

The release version is 0.2.0. You can download the source code from [OAID/caffeOnACL](https://github.com/OAID/caffeOnACL)

The release is verified on 64bits ARMv8 processor<br>

* Hardware platform : Rockchip RK3399 (firefly RK3399 board)<br>

* Software platform : Ubuntu 16.04<br>

* ConvolutionLayer

* PoolingLayer

* LRNLayer

* ReLULayer

* SigmoidLayer

* SoftmaxLayer

* TanHLayer

* AbsValLayer

* BNLLLayer

* InnerProductLayer

There are some compatibility issues between ACL and caffe Layers, we bypass it to Caffe's original layer class as the workaround solution for the below issues

* Normalization in-channel issue

* Tanh issue

* Even Kernel size

* Softmax supporting multi-dimension issue

* Group issue

* Performance need be fine turned in the future

The caffe based version is `793bd96351749cb8df16f1581baf3e7d8036ac37`.

Makefile.config.acl

cmake/Modules/FindACL.cmake

examples/cpp_classification/classification_profiling.cpp

examples/cpp_classification/classification_profiling_gpu.cpp

include/caffe/acl_layer.hpp

include/caffe/layers/acl_absval_layer.hpp

include/caffe/layers/acl_base_activation_layer.hpp

include/caffe/layers/acl_bnll_layer.hpp

include/caffe/layers/acl_conv_layer.hpp

include/caffe/layers/acl_inner_product_layer.hpp

include/caffe/layers/acl_lrn_layer.hpp

include/caffe/layers/acl_pooling_layer.hpp

include/caffe/layers/acl_relu_layer.hpp

include/caffe/layers/acl_sigmoid_layer.hpp

include/caffe/layers/acl_softmax_layer.hpp

include/caffe/layers/acl_tanh_layer.hpp

models/SqueezeNet/README.md

models/SqueezeNet/SqueezeNet_v1.1/squeezenet.1.1.deploy.prototxt

src/caffe/acl_layer.cpp

src/caffe/layers/acl_absval_layer.cpp

src/caffe/layers/acl_base_activation_layer.cpp

src/caffe/layers/acl_bnll_layer.cpp

src/caffe/layers/acl_conv_layer.cpp

src/caffe/layers/acl_inner_product_layer.cpp

src/caffe/layers/acl_lrn_layer.cpp

src/caffe/layers/acl_pooling_layer.cpp

src/caffe/layers/acl_relu_layer.cpp

src/caffe/layers/acl_sigmoid_layer.cpp

src/caffe/layers/acl_softmax_layer.cpp

src/caffe/layers/acl_tanh_layer.cpp

unit_tests/Makefile

unit_tests/pmu.c

unit_tests/pmu.h

unit_tests/prof_convolution_layer.cpp

unit_tests/sgemm.cpp

unit_tests/test.cpp

unit_tests/test_caffe_main.cpp

unit_tests/test_common.cpp

unit_tests/test_convolution_layer.cpp

unit_tests/test_fail.cpp

unit_tests/test_inner_product_layer.cpp

unit_tests/test_lrn_layer.cpp

unit_tests/test_neuron_layer.cpp

unit_tests/test_pooling_layer.cpp

unit_tests/test_softmax_layer.cpp

unit_tests/testbed.c

unit_tests/testbed.h

Makefile

cmake/Dependencies.cmake

include/caffe/caffe.hpp

include/caffe/common.hpp

include/caffe/layer.hpp

include/caffe/util/device_alternate.hpp

include/caffe/util/hdf5.hpp

src/caffe/common.cpp

src/caffe/layer.cpp

src/caffe/layer_factory.cpp

src/caffe/layers/absval_layer.cpp

src/caffe/layers/bnll_layer.cpp

src/caffe/layers/hdf5_data_layer.cpp

src/caffe/layers/hdf5_data_layer.cu

src/caffe/layers/hdf5_output_layer.cpp

src/caffe/layers/hdf5_output_layer.cu

src/caffe/layers/inner_product_layer.cpp

src/caffe/net.cpp

src/caffe/solvers/sgd_solver.cpp

src/caffe/syncedmem.cpp

src/caffe/test/test_hdf5_output_layer.cpp

src/caffe/test/test_hdf5data_layer.cpp

src/caffe/util/hdf5.cpp

src/caffe/util/math_functions.cpp

Encounter any issue, please report on [issue report](https://github.com/OAID/caffeOnACL/issues). Issue report should contain the following information :

* The exact description of the steps that are needed to reproduce the issue

* The exact description of what happens and what you think is wrong

Download the model files from https://github.com/DeepScale/SqueezeNet/tree/master/SqueezeNet_v1.1

The model architecture file for deploying is:

SqueezeNet_v1.1/squeezenet.1.1.deploy.prototxt