Skip to content

Latest commit

 

History

History
122 lines (91 loc) · 3.81 KB

INSTALL_INSTRUCTIONS.md

File metadata and controls

122 lines (91 loc) · 3.81 KB

DELF installation

Tensorflow

TensorFlow 2.1 Python 3.6

For detailed steps to install Tensorflow, follow the Tensorflow installation instructions. A typical user can install Tensorflow using one of the following commands:

# For CPU:
pip3 install 'tensorflow'
# For GPU:
pip3 install 'tensorflow-gpu'

TF-Slim

Note: currently, we need to install the latest version from source, to avoid using previous versions which relied on tf.contrib (which is now deprecated).

git clone [email protected]:google-research/tf-slim.git
cd tf-slim
pip3 install .

Note that these commands assume you are cloning using SSH. If you are using HTTPS instead, use git clone https://github.com/google-research/tf-slim.git instead. See this link for more information.

Protobuf

The DELF library uses protobuf (the python version) to configure feature extraction and its format. You will need the protoc compiler, version >= 3.3. The easiest way to get it is to download directly. For Linux, this can be done as (see here for other platforms):

wget https://github.com/google/protobuf/releases/download/v3.3.0/protoc-3.3.0-linux-x86_64.zip
unzip protoc-3.3.0-linux-x86_64.zip
PATH_TO_PROTOC=`pwd`

Python dependencies

Install python library dependencies:

pip3 install matplotlib numpy scikit-image scipy
sudo apt-get install python3-tk

tensorflow/models

Now, clone tensorflow/models, and install required libraries: (note that the object_detection library requires you to add tensorflow/models/research/ to your PYTHONPATH, as instructed here)

git clone [email protected]:tensorflow/models.git

# Setup the object_detection module by editing PYTHONPATH.
cd ..
# From tensorflow/models/research/
export PYTHONPATH=$PYTHONPATH:`pwd`

Note that these commands assume you are cloning using SSH. If you are using HTTPS instead, use git clone https://github.com/tensorflow/models.git instead. See this link for more information.

Then, compile DELF's protobufs. Use PATH_TO_PROTOC as the directory where you downloaded the protoc compiler.

# From tensorflow/models/research/delf/
${PATH_TO_PROTOC?}/bin/protoc delf/protos/*.proto --python_out=.

Finally, install the DELF package. This may also install some other dependencies under the hood.

# From tensorflow/models/research/delf/
pip3 install -e . # Install "delf" package.

At this point, running

python3 -c 'import delf'

should just return without complaints. This indicates that the DELF package is loaded successfully.

Troubleshooting

pip3 install

Issues might be observed if using pip3 install with -e option (editable mode). You may try out to simply remove the -e from the commands above. Also, depending on your machine setup, you might need to run the sudo pip3 install command, that is with a sudo at the beginning.

Cloning github repositories

The default commands above assume you are cloning using SSH. If you are using HTTPS instead, use for example git clone https://github.com/tensorflow/models.git instead of git clone [email protected]:tensorflow/models.git. See this link for more information.