Dependency:
Python 3.4.3
Tensorflow> 1.0.0, tqdm, cv2, exifread, skimage, glob
1. Install tensorflow:Https://www.tensorflow.org/install/pip? Lang = python3
1. Install the python Development Environment
sudo apt updatesudo apt install python3-dev python3-pipsudo pip3 install -U virtualenv
2. Create a virtual environment (Python virtual environment is used to isolate package installation from the system .)
Virtualenv -- System-site-packages-P python3./Venv# Create a directory named venv to save the Virtual Environment
Source./Venv/Bin/activate # activate the Virtual Environment
# After activation, add (venv) to the shell prompt CD)
PIP install -- upgrade Pip # upgradingpip
PIP list #how packages installed within the virtual environment
deactivate # exit virtualenv
3. Install tensflow using Pip
PIP install tensorflow
Import tensorflow as TF; print (Tf. _ version _) "# output 1.11.0
Https://www.tensorflow.org/install/gpu this is the link address:
Only after installationCuda toolkit7.5 andIn 64-bit Ubuntu of cudnn V4, you can use Pip to install tensflow that supports GPU. Otherwise, you must use the source code for installation.
To support GPU, installCuda Toolkit(V> 7.0) andCudnn(V> = 2)
Tensorflow only supports GPU with a computing capacity greater than 3.0.
--------- Tensflow has been installed.
2. Install tqdmHttps://github.com/tqdm/tqdmthis is an introduction
pip install tqdm
Version 4.27.0 is installed.
3. Install cv2
pip install opencv-python
The installed version is: opencv-python-3.4.3.18
4. Install exifreadHttps://pypi.org/project/ExifRead/
pip install ExifRead
The installed version is: (2.1.2)
5. Install scikit-ImageHttps://scikit-image.org/
pip install -U scikit-image
6. Install globHttps://pypi.org/project/glob2/
Glob can be used to find the file path name that complies with specific rules. Only three matching characters are used for searching files :"*","? "," []"
pip install glob2
The dependency is installed here.
----------------------------------------------- Split line ----------------------- The following is the usage)
1,Get the dataset, first line download https://www.dropbox.com/s/l7a8zvni6ia5f9g/datasets.tar.gz? DL = 0
Decompress the downloaded data to the./data directory.
2,Run code
CD. /script # Switch the current path. /Batch \ _ run_train.sh # Run the script to generate the transformed patch and train the model. /Batch \ _ run_test.sh # extract local feature point. /Batch \ _ run_eval.sh # evaluate the performance
Learning discriminative and transformation covariant local feature detectors lab environment setup detailed process