Automated Grading of Baseball Trading Cards using Convolutional Neural Networks

Authors:

• Brian Miller ([email protected])
• Sean Wendlandt ([email protected])

Project Overview:

The condition of a trading card significantly influences its worth. However, the current methods to assess a card's condition are insufficient. Non-professionals often lack the accuracy required for card grading, and obtaining professional evaluations is too expensive and time-consuming.

This project aims to automate the card grading process using machine learning. Our neural network utilizes the pretained model ResNet-50 combined with custom classification and output layers.

Data Preprocessing Steps:

• Web scraping to obtain data
• Manual removal of innacurate data
• Dataset balancing
• Automated removal of PSA label
• Standardize photo dimensions for CNN

Description of CNN:

• ResNet-50 feeds into three fully connected ReLU dense layers [1024, 512, 128]
• The dense layers feed into a single softmax prediction layer

Two rounds of training were done:

• First Round: 50 epochs, freeze first 30 layers of ResNet-50
• Second Round: 50 epochs, unfreeze all layers

Data:

All 10k cards were downloaded from Collectors.com

Description of Files: