Climbing Hold Classification
Developed transferred learning model for Detectron2 that was specifically trained to classify climbing holds
For my Computer Vision course in 2021, I developed a model that was used to classify climbing holds. Climbing holds, while having a wide range of shapes and sizes, are classified into specific categories. The goal was to try to do transferred learning on the Detectron2 model to classify holds into distinct categories. Training to detect very disparate features to classify for the same category proved to be difficult for the model.