MLRF official web page

Welcome to MLRF official web page

This web page is the official web page for the MLRF course taught at EPITA. Students should find listed here (almost) all the resources they need for this course.

Teachers: Joseph Chazalon (lectures + practice), Olivier Ricou (practice).

List of home and graded work:

1. First session about global descriptors:
   • (nothing — IML project)
2. Local keypoint detectors and descriptors:
   • Quiz on Moodle
3. Local feature detectors and descriptors, Matching descriptors, Perspective transform estimation:
   • Quiz on Moodle
4. Content based image retrieval and Evaluation
   • Quiz on Moodle
   • Lab assignment: submit notebook and dataset results on Moodle
5. Image classification and Evaluation
   • Quiz on Moodle
   • Lab assignment: submit notebook and dataset results on Moodle
6. Image classification, Image Segmentation, Research problems, Conclusions
   • Quiz on Moodle
   • Lab assignment: submit notebook and dataset results on Moodle

Tools we are using

We will try to use the right tool for the right job. Here are the tools and their job(s).

• This website: to share files with you
• Moodle:
  • for graded tests
  • to submit results for practice sessions 4, 5 and 6
  • for the final feedback about the course
• email and Teams: on rare occasions for important communication about the course

Setting up a programming environment on your computer

You need to set up a programming environment on your computer to complete this course. We tried to make this as easy as possible.

We use Jupyter, a notebook interface for scientific Python. We provide you with notebooks that you have to edit to complete each practice session.

To get a working environment, you first need to install a recent (3.7+) Python environment. Then, you need to install some packages to get all the necessary tools We recommend using pip, the Python package installer, as follows;

pip install --user \
    jupyter \
    matplotlib \
    numpy \
    opencv-contrib-python-headless \
    scikit-image \
    scikit-learn

You may need to edit your $PATH and add $HOME/.local/bin to the list of paths.

Then, you should be able to navigate to a directory containing Jupyter Notebooks (.ipynb files) and start a Jupyter server using

jupyter notebook

This should automatically launch a browser and let you interact with notebooks. We recommend that you read those two resources to get Jupyter’s basics:

• Jupyter Documentation — Running the Notebook
• Jupyter Notebook Overview

Using EPITA lab computers

Course resources are available on the AFS at the following mount point: /afs/cri.epita.fr/resources/teach/bigdata/mlrf21

To activate necessary packages on Nix, you may need to use our custom derivation:

wget https://www.lrde.epita.fr/~jchazalo/teaching/MLRF/202204_IMAGE_S8/static/mlrf-shell.nix
nix-shell mlrf-shell.nix
# then, you can launch jupyter-notebook

To save your Jupyter and IPython configurations you may want to use the following commands:

mkdir -p ~/afs/.confs/jupyter/
mkdir -p ~/afs/.confs/ipython/
ln -sf ~/afs/.confs/jupyter/ ~/.jupyter
ln -sf ~/afs/.confs/jupyter/ ~/.ipython

Resources

Remarks:

• This list will be updated along the course.
• Elements as listed in the order you should use or complete them.
• For practice sessions, please see the README file in the student resource archive for instructions.
• Due to copyright restrictions, links to datasets will be sent using private group chat.