Intel Image Classification Student: Le Dinh Nam Student ID: V202000210
Overview of the project: I want to build a model that can classify different natural scenes into 6 categories: buildings, streets, glaciers, mountains, forests, sea.
This directory includes:
- Source code of the project: Intel Image Classification (ipynb file)
- Report of the project (pdf file)
- The data sets (seg_pred, seg_train, seg_test)
Libraries used:
- Torch
- Torchvision
- Numpy
- Matplotlib
Description:
The project is part of the Introduction to Machine Learning course. This project provides the Intel dataset of images with 6 labels and a test set which will be evaluated after running the resnet50 model. The results are surprising with an accuracy of nearly 93%, higher than most of the models used on the Intel platform at that time.
References:
- https://www.youtube.com/watch?v=GIsg-ZUy0MY&t=21364s - Pytorch tutorial
- https://www.kaggle.com/puneet6060/intel-image-classification - Data from Kaggle
- https://pytorch.org/tutorials/beginner/finetuning_torchvision_models_tutorial.html - Finetuning Torch Vision model