People

• YoungJeansKR
• haen-su
• ch9eri

Language / IDE

• 
• 
• 
• 
• 

Dataset

• 2140 images
• 2140 xml files

Model

• CNN (Conv - ReLU - Pooling - Conv - ReLU - Pooling - Affine - ReLU - Affine - Sigmoid )

Optimizer

• SGD, Momentum, Adam
• Lr = 0.001, 0.01, 0.1
• Batch size = 32, 64
• Epochs = 20, 50, 100

Implementation Example Code

Result

• Train Accuracy : 89.91%
• Test Accuracy : 77.32%