ImageNet100

2020-2st 전자공학종합설계 Project

1. Contributor

• 양재원

2. Version

• python3.7
• torch1.7 and torchvision 0.6
• Cuda >=10.2
• Ubuntu 18.04
• Multiple GPU: Nvidia Titan X

3. Train

• Must to do: You need to unzip the ImageNet100.zip into the project imageNet-100 dir. Train and test should be unziped.
• command: python train.py (Check config.py for datapath and hyper-parameter settings)

4. Test

• Command: python test.py (Check config.py for datapath and hyper-parameter settings)
• output: model accuracy

5. Best Model Test

• Unzip the Best_model.zip and move the models to the output/result dir
• Check config.py
• Run python test.py