3D Model Search Engine

06

This multimedia retrieval pipeline was developed for the Multimedia Retrieval course at Utrecht University. It functions as a 3D search engine that accepts a 3D model as input and returns the top 10 most similar models from the database. Each model in the database undergoes normalization, which includes fitting it into a unit box, standardizing the number of vertices, and orienting it using Principal Component Analysis. Models are then compared based on specific descriptors and histograms.

Although the previous version allowed users to upload a model and query the database, the current version lacks a backend for preprocessing uploaded models. Nevertheless, it remains an interesting demo for visualizing models and identifying similar ones.