Download the miniforge3 distribution for your Operating System (Windows, macOS or Linux):
https://github.com/conda-forge/miniforge#miniforge3
miniforge is a conda installation that uses the packages from the conda-forge installation by default. We recommand using this instead of Anaconda/miniconda3 because conda-forge tends to be more up-to-date and support more platforms than the default channel of Anaconda/miniconda. However both might work for this class.
Optional: feel free to create a dedicated conda environment for this class if you don't want to mess with Python dependencies needed for other classes or projects:
conda create -n dlclass python=3.9
conda activate dlclass
Install or update the following packages with the conda command:
conda install -y tensorflow scikit-learn pandas jupyterlab matplotlib-base
conda install -y h5py pillow scikit-image lxml pip ipykernel
Check that you can import tensorflow with the python from anaconda:
python -c "import tensorflow as tf; print(tf.__version__)"
2.6.0
Note that any tensorflow version from 2.0.0 should work for this class.
If you have several installations of Python on your system (virtualenv, conda environments...), it can be confusing to select the correct Python environment from the jupyter interface. You can name this environment for instance "dlclass" and reference it as a Jupyter kernel:
python -m ipykernel install --user --name dlclass --display-name dlclass
Ideally: create a new jupyter notebook and check that you can import the numpy, matplotlib, tensorflow modules.
To take pictures with the webcam we will also need opencv-python:
python -m pip install opencv-python
If your laptop does not have a webcam or if opencv does not work, don't worry this is not mandatory.
In a console check the installation location of the conda command in your PATH:
conda info
Read the output of that command to verify that your conda command is installed where you expect it to be. If it's not the case, you might want to change the order in your PATH variable (for instance in your $HOME/.bashrc or $HOME/.zshrc file on Linux and macOS) accordingly.
Check that the pip command in your PATH is the one installed by conda:
pip show pip
and check that it matches:
python -m pip show pip
In particular, look at the "Location:" line of pip is a subfolder of the "environment root:" line from "conda info".
If you cannot solve your installations without the help of fellow students, please feel free to contact instructors (preferably in the slack channel of the class) with the outputs of the previous command and include the full error messages along with the name and version of your operating system.