/Memoir

Python package to colorize black and white videos.

Primary LanguagePythonMIT LicenseMIT

Dataset

The videos used for generating the dataset are available here.

Documentation

The documentation is available here.

Installation

Fork the repository.

Clone the forked repository.

$ git clone <link_to_forked_repository>

Change the directory to Memoir.

$ cd /path/to/Memoir/

Install the dependencies.

For Python (virtual) environment:

$ pip3 install numpy
$ pip3 install scikit-image
$ pip3 install opencv-python
$ pip3 install tensorflow==1.14 (cpu)
$ pip3 install tensorflow-gpu==1.14 (gpu)**

** You will need CUDA and cuDNN libraries to run tensorflow on GPU.

Building Tensorflow 1.14 with GPU Support and TensorRT on Ubuntu 16.04 LTS

Building Tensorflow 2.0 with GPU support and TensorRT on Ubuntu 18.04 LTS

For Conda environment:

Either use anaconda-navigator to install the dependencies, including tensorflow-gpu==1.14. It will automatically take care of CUDA and cuDNN. Or you could just duplicate the developer's environment using:

conda env create -f requirements.yml

Install the package Memoir:

For both the environments.

(Memoir)$ pip install .

to install in editable (developer) mode:

(Memoir)$ pip install -e .

To uninstall Memoir,

$ pip uninstall -y Memoir