Note: To use the scripts, you must add the tool dataset in th same folder. You can find it at : tools.tar.gz .It contains everything related to this dataset.
Image classificaion algorithm for a tool dataset
Hugo RICHARD
Ecole Nationale supérieure des Arts et Métiers
8 Boulevard Louis XIV
59800 LILLE
FRANCE
Tel. (+33) 760964267
email: [email protected]
This repository contains :
+ Jupyter Notebooks, that are useful to first understand the code :
- Extract features wit VGG16.ipynb
- Feature Visualization (PCA and t-SNE).ipynb
- NN Classification.ipynb
- Stock images into numpy array.ipynb
- SVM Classification.ipynb
+ saved numpy arrays (only useful inside th code) :
- features.npy
- other numpy arrays in the features_fc1 and labels folder
+ Python scripts :
- Extract features wit VGG16.py
- Feature Visualization (PCA and t-SNE).py
- NN Classification.py
- Stock images into numpy array.py
- SVM Classification.py
First, with the 'Stock images into numpy array.ipynb' we transform the database from images in subfolders into a usefull numpy array.
Then with the 'Extract features wit VGG16.ipynb', we use a pretrained CNN to extract features from the images.
Now w want to visualize the features with PCA and t-SNE in 'Feature Visualization (PCA and t-SNE).ipynb'
The next step is to manage to classify these features. First we try with a SVM in 'SVM Classification.ipynb', then with a Neural Network in 'NN Classification.ipynb'.
I am currently working on classifying the features with siamese networks, but before posting anything I still have to work on it.