The gqcnn Python package is for training and analysis of Grasp Quality Convolutional Neural Networks (GQ-CNNs). It is part of the ongoing Dexterity-Network (Dex-Net) project created and maintained by the AUTOLAB at UC Berkeley.
Any system may work if it can be installed tensorflow 2. CUDA and CuDNN versions depend on your specific GPUs.
OS: Ubuntu 20.04
GPU: NVIDIA RTX A6000 48GB
CUDA: 11.2
cuDNN: ??
Nvidia driver: 470.129.06
python: 3.8.10
OS: Rocky Linux (ABCI 2.0 V-node)
GPU: NVIDIA V100 for NVLink 16GiB HBM2 x4
gcc: 12.2 (gcc/12.2.0)
python: 3.10.10 (python/3.10/3.10.10)
CUDA: 11.8 (cuda/11.8/11.8.0)
cuDNN: 8.6.0 (cudnn/8.6/8.6.0)
Versions (the original script is written in tensoflow 1.15, but it uses tensorflow 2.12 and tensorflow.compat.v1)
Package Version Editable project location \
---------------------------- --------- ------------------------------------
absl-py 1.4.0
astunparse 1.6.3
attrs 23.1.0
autolab-core 1.1.1
autolab-perception 1.0.0
cachetools 5.3.0
certifi 2023.5.7
chardet 5.1.0
charset-normalizer 3.1.0
colorlog 6.7.0
contourpy 1.0.7
cycler 0.11.0
dill 0.3.6
ffmpeg-python 0.2.0
flatbuffers 23.5.9
fonttools 4.39.4
freetype-py 2.4.0
future 0.18.3
gast 0.4.0
google-auth 2.18.0
google-auth-oauthlib 1.0.0
google-pasta 0.2.0
GPUtil 1.4.0
gqcnn 1.3.0
grpcio 1.54.2
h5py 3.8.0
idna 3.4
imageio 2.28.1
jax 0.4.10
joblib 1.2.0
jsonschema 4.18.0a6
jsonschema-specifications 2023.3.6
keras 2.12.0
kiwisolver 1.4.4
lazy_loader 0.2
libclang 16.0.0
lxml 4.9.2
mapbox-earcut 1.0.1
Markdown 3.4.3
MarkupSafe 2.1.2
matplotlib 3.7.1
ml-dtypes 0.1.0
mpmath 1.3.0
multiprocess 0.70.14
networkx 3.1
numpy 1.23.5
oauthlib 3.2.2
opencv-python 4.7.0.72
opt-einsum 3.3.0
packaging 23.1
Pillow 9.5.0
pip 22.3.1
protobuf 4.23.0
psutil 5.9.5
pyasn1 0.5.0
pyasn1-modules 0.3.0
pycollada 0.7.2
pyglet 2.0.7
PyOpenGL 3.1.0
pyparsing 3.1.0b1
pyrender 0.1.45
pyserial 3.5
python-dateutil 2.8.2
PyWavelets 1.4.1
referencing 0.28.1
requests 2.30.0
requests-oauthlib 1.3.1
rpds-py 0.7.1
rsa 4.9
Rtree 1.0.1
ruamel.std.argparse 0.8.3
ruamel.yaml 0.2
ruamel.yaml.clib 0.2.7
scikit-image 0.21.0rc0
scikit-learn 1.2.2
scipy 1.10.1
setproctitle 1.3.2
setuptools 51.1.2
shapely 2.0.1
six 1.16.0
svg.path 6.3
sympy 1.12
tensorboard 2.12.3
tensorboard-data-server 0.7.0
tensorflow 2.12.0
tensorflow-estimator 2.12.0
tensorflow-io-gcs-filesystem 0.32.0
termcolor 2.3.0
threadpoolctl 3.1.0
tifffile 2023.4.12
trimesh 3.21.6
typing_extensions 4.5.0
urllib3 1.26.15
visualization 1.0.0
Werkzeug 2.3.4
wheel 0.40.0
wrapt 1.14.1
xxhash 3.2.0
Please see the docs for installation and usage instructions.
If you only use in python, please do the followings
cd /path/to/gqcnn
pip install .If you use in ROS, see the docs. But in this repository, the ROS implementation is not confirmed in python3.
In pip installation, something problem may happen in depended packages. Berkeley-AUTOLAB specific packages should be installed.
- autolab-core
- autolab-perception
- visualization
The installation is following to Dexnet install docs
# go to the root directory that you cloned the package
cd /path/to/gqcnn
mkdir deps
cd deps
git clone https://github.com/BerkeleyAutomation/autolab_core.git
git clone https://github.com/BerkeleyAutomation/perception.git
git clone https://github.com/k-makihara/visualization.git
# install autolab_core
cd autolab_core
python setup.py develop
cd ..
# install perception
cd perception
python setup.py develop
cd ..
# install visualization
cd visualization
python setup.py develop
cd ..
and, please do again in the docs
cd /path/to/gqcnn
pip install .In these installations, if you may meet additional errors, please install this package from apt and pip. Not all packages are necessary, so it's better to select and install only the required packages.
# install apt deps
sudo apt-get install cmake libvtk5-dev python-vtk python-sip python-qt4 libosmesa6-dev meshlab libhdf5-dev
# install pip deps
pip install numpy scipy scikit-learn scikit-image opencv-python pyassimp tensorflow h5py mayavi matplotlib catkin_pkg multiprocess dill cvxopt ipython pillow pyhull setproctitle trimesh
Singularity is confirmed, use this Dockerfile and compile, but local installation is better to use because it may something problem still remained
FROM nvcr.io/nvidia/tensorflow:20.01-tf1-py2
ENV DEBIAN_FRONTEND=noninteractive
ENV TZ=Asia/Tokyo
RUN ln -snf /usr/share/zoneinfo/$TZ /etc/localtime && echo $TZ > /etc/timezone
ENV DISPLAY host.docker.internal:0.0
RUN apt-get -y update && apt-get -y upgrade && \
apt-get install -y software-properties-common && \
add-apt-repository -y ppa:openscad/releases && \
apt-get -y update && \
apt-get -y install \
x11-apps \
build-essential \
cmake \
wget \
git \
nano \
curl \
vim \
apt-utils \
python-dev \
python-pip \
python3-dev \
python3-pip \
libglib2.0-0 \
xvfb \
libgl1-mesa-dev \
libxkbcommon-x11-0 \
libavcodec-dev \
libavformat-dev \
libswscale-dev \
qt5-default \
libqt5widgets5 \
unity-gtk-module-common \
unity-gtk3-module \
xorg \
libgl1-mesa-glx \
mesa-utils \
language-pack-en-base \
libglew-dev \
libcanberra-gtk-module \
libcanberra-gtk3-module \
bash \
gcc \
gfortran \
g++ \
make \
file \
unzip \
python-sip \
python-qt4 \
libosmesa6-dev \
meshlab \
libhdf5-dev \
freeglut3 \
freeglut3-dev \
openscad \
blender \
python-tk \
ffmpeg \
default-jdk
RUN mkdir -p /home/co
# Install libspatialindex
RUN cd /home/co && \
wget https://download.osgeo.org/libspatialindex/spatialindex-src-1.7.0.tar.gz && \
tar -xzf spatialindex-src-1.7.0.tar.gz && \
cd spatialindex-src-1.7.0 && \
./configure && \
make && \
make install && \
ldconfig
# Install newest pip versions
#RUN curl https://bootstrap.pypa.io/get-pip.py -o get-pip.py && \
# python get-pip.py && \
# python3 get-pip.py && \
# pip2 install --upgrade setuptools && \
# pip3 install --upgrade setuptools
# Install python2 dependencies (largely very specific to get dexnet to actually run).
RUN pip2 install numpy==1.16.5 scipy==1.2.2 scikit-learn==0.20.4 scikit-image==0.14.5 opencv-python==4.1.2.30 pyassimp tensorflow==2.0.0 h5py networkx==2.2 tornado==3.2
RUN pip2 install imageio==2.6.1
RUN pip2 install vtk
RUN pip2 install traitsui==6.1.3 pyface==6.1.2 mayavi==4.5.0
RUN pip2 install matplotlib==2.2.4 multiprocess dill ipython==5.8.0 Pillow==6.2.1 setproctitle trimesh
RUN pip2 install pyhull
RUN pip2 install ffmpeg-python
RUN pip2 install pathlib2 pathlib
RUN pip2 install joblib==0.13.1 matplotlib==2.2.0 ruamel.yaml==0.16.0
RUN pip2 install cvxopt==1.1.9
RUN pip2 install colorlog==5.0.1
RUN pip2 install shapely
# Install Dex-Net and dependencies.
RUN mkdir -p /home/co/dexnet && \
cd /home/co/dexnet && \
wget https://github.com/dougsm/egad/releases/download/v0.1/dexnet_deployment.zip --no-check-certificate && \
unzip dexnet_deployment.zip && \
cd deps && \
# install SDFGen
cd SDFGen && \
sh install.sh && \
cd .. && \
# install Boost.NumPy
cd Boost.NumPy && \
sh install.sh && \
cd .. && \
# return to dex-net directory
# install meshpy
cd meshpy && \
python setup.py develop && \
cd .. && \
# install all Berkeley AUTOLAB modules
# autolab_core
cd autolab_core && \
python setup.py develop && \
cd .. && \
# perception
cd perception && \
python setup.py develop && \
cd .. && \
# visualization
cd visualization && \
python setup.py develop && \
cd .. && \
# gqcnn
cd gqcnn && \
pip install . && \
cd .. && \
# dexnet
cd .. && \
python setup.py develop
RUN cd /home/co && \
git clone https://github.com/k-makihara/openrave-installation.git && \
cd openrave-installation && \
./install-dependencies.sh && \
./install-osg.sh && \
./install-fcl.sh && \
./install-openrave.sh
If you use any part of this code in a publication, please cite the appropriate Dex-Net publication.
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent