HYU-AUE8088, Understanding and Utilizing Deep Learning

PA #1. Image Classification

Files

├── README.md
├── requirements.txt
├── src
│   ├── __init__.py
│   ├── config.py
│   ├── dataset.py
│   ├── metric.py
│   ├── network.py
│   └── util.py
├── test.py
└── train.py

0. Preparation

Setup virtual environment

• Create python virtual environment

$ python3 -m venv venv/aue8088
$ source venv/aue8088/bin/activate

• Check whether the virtual environment set properly : The result should end with venv/aue8088/bin/python.

$ which python

• Install required packages

$ pip install -r requirements.txt

Wandb setup

• Login

$ wandb login

• Specify your Wandb entity

$ echo "export WANDB_ENTITY={YOUR_WANDB_ENTITY}" >> ~/.bashrc
$ source ~/.bashrc

1. [TODO] Evaluation metric

Finish MyAccuracy class (src/metric.py)

• Please complete this function to measure accuracy of the prediction

Implement MyF1Score class (src/metric.py)

• Please write MyF1Score class from scratch
  • Calculate per-class F-1 score in a one-vs-rest manner
• Apply this new metric (hint: update src/network.py)

2. [TODO] Train models

• Try different settings (src/config.py)

$ python train.py

3. [TODO] Toward state-of-the-art

• How to improve performance more?
  • Find state-of-the-art method/model(paper) on TinyImageNet-200 dataset
  • Check difference between baseline and state-of-the-art
  • Apply missing stuff in the baseline