"Introduction to machine learning" project - Polytech ET4 IT - Noted for S8 (2020-2021)
The subject has been chosen by the students. The unsupervised learning model too..
All remaining bugs are listed here.
These instructions will get you a copy of the project up and running on your local machine for development purpose.
Things you need to install the project :
- Jupyter
- Python 3
- Scikit-Learn library
- Scikit-Image library
- NumPy library
- Matplotlib library
Here are some instructions on how to get the development env running.
First, clone this repository with the following command :
$ git clone https://github.com/adepreis/BzzML
Then, after moving into the /BzzML folder just created, start Jupyter using :
$ jupyter notebook
or
> jupyter-notebook.exe
if you are a Windows user.
This should open up your browser (if it doesn't, visit 127.0.0.1:8888), and you should see Jupyter's tree view, with the contents of the current directory.
If you are not familiar with Jupyter Notebooks, the most recent release includes the corresponding .py files.
After being pre-proceeded (reduced 15 times and "sobel-filtered"), the dataset is divided in training and validation sets before being passed to a PCA (Principal Component Analysis) and then to a Gaussian Naive Bayesian model.
| Gaussian Naive Bayesian | SVC (Support Vector Classifier) |
|---|---|
 |
 |
As you can see, we obtained even better results with a Support Vector Classifier. Some work has also been done on the SVC parameters :
- Lucas B. - @0xWryth
- Antonin D. - @adepreis
In the /data folder, you can find the /image folder where you should place the dataset images.
The used image dataset has been shared by Ivan Felipe Rodriguez under Honey Bee pollen kaggle repository in the framework of the following publication :
Ivan Rodriguez, Rémi Mégret, Edgar Acuña, José Agosto, Tugrul Giray. Recognition of pollen-bearing bees from Video using Convolutional Neural Network, IEEE Winter Conf. on Applications of Computer Vision, 2018, Lake Tahoe, NV. https://doi.org/10.1109/WACV.2018.00041
In the /doc folder, you can find a brief report that explains the design choices and contains result screenshots.