Comparison of Deep Learning/ CNN for image prediction
Project Details
- Project Type: Machine learning
- Tools: Python, Jupyter Notebook with Tensorflow library
- Duration: 2 weeks
- Team: Duo project with Antoine Banchet
- Code: Github noebackert/Machine-learningEMSE
- Report: Download report
Introduction
The goal of this project was to understand the basis of machine learning and deep learning. We compared the performance of different models for image prediction.
Project Overview
This project encompasses:
- Comparing various models for image prediction
- Training and testing these models
- Evaluating their performance
- Gaining hands-on experience
Project Description
Our project aimed to compare the performance of different models for image prediction. We used the CIFAR-3 dataset, which contains 18,000 32x32 RGB images in 3 different classes. The dataset is divided into 14,400 training images and 3,600 testing images.
Models
We compared the performance of the following models:
- Logistic Regression
- Gaussian Naive Bayes Classifier
- Deep Learning: MLP
- Convolutional Neural Network (CNN)
Results
Our results showed that the CNN model outperformed the other models. It achieved an accuracy of 88% on the test set, while the MLP model achieved 75% accuracy. The logistic regression and Gaussian Naive Bayes models achieved 60% and 61% accuracy, respectively.