Machine Learning-Based Texture Classification Tool
Beta Version
About the project
This website houses a tool developed by the team at the Yale IPCH Lens Media Lab for the use of characterizing texture data taken from various samples of photographic paper. The tool primarily uses a characterization model known as the Semi-Supervised Triplet Loss Neural Network for Assessing Paper Texture Similarity, which was originally developed within the ASPECT Lab at Western Washington University.
The primary use for this tool is to create a visual representation of a paper’s texture, which can range from smooth to rough, and with a wide degree of variation in-between. This tool aims to identify and specify that in-between, and also match pre-loaded paper textures in our database with potentially those uploaded by researchers. This tool will hopefully aid conservators, researchers, and scholars interested in the history of photography, to trace their texture profiles and develop new points of understanding photographic paper collections.
