The convolutional neural network (designated by the acronym CNN) is one of the most efficient models for performing image recognition and classification, object detection and face recognition.
GoFocus members :
- Meriem AMERAOUI
- Dounia BELABIOD
- Jihene BOUHLEL
- Bahaa Eddine NIL
Python 3 is used during development and the following libraries are required to run the code provided in the notebook :
- NumPy = 1.17.2
- Matplotlib = 3.1.1
- Keras = 2.3.1
- TensorFlow = 2.2.0
The classes folder contains all the classes we've implemented.
conv.py: A convolution layer using 3x3 filters.convolutionalneuralnetwork.py: Main class of our network.dense.py: A standard fully-connected layer with softmax activation.dropout.py: A dropout layer to avoid the model of over learning.flatten.py: A flattening layer.maxpool.py: A max pooling layer using a pool size of 2.relu.py: Rectified linear units activation function.
We have 3 notebooks in this project :
ConvolutionalNeuralNetwok.ipynb: Our main file where we display the data and create the model to classify images.DataGenerator.ipynb: In this file we generate more data from the same image.KerasModel.ipynb: Creation of the Keras model to classify images.
The requirements.txt file lists all the Python libraries on which the notebooks of this project depend, you can install them using :
pip install -r requirements.txt
The ressouces folder contains all the secondary files necessary for the project (Gantt diagram, report, video presentation and slides).
Deep learning, convolutional neural network.