CAD Indexing and Retrieval Using HOOPS Embeddings

These notebooks demonstrate how to compute CAD shape embeddings, save a FAISS index, and retrieve geometrically similar parts from a CAD query file.

Follow Install HOOPS AI for environment and license setup. The indexing notebook uses the pre-trained SIGNAL model; training a custom model with Preparing Data and Training Custom HOOPS Embeddings is optional.

Work through the notebooks in this order:

1. demo_HOOPS_Embeddings_indexing.ipynb loads a CAD dataset, computes embeddings in batch, builds a FAISS index, and saves it to disk.

2. demo_HOOPS_Embeddings_retrieval.ipynb loads that index and searches for similar parts using a query CAD file.

The code and resources for this tutorial can be found in the HOOPS-AI-Tutorials GitHub repository.

Info

From the repository root, run jupyter lab embeddings_pipeline/demo_HOOPS_Embeddings_indexing.ipynb to begin the indexing workflow.

• Indexing a CAD Dataset with HOOPS Embeddings
• CAD Similarity Search Using HOOPS Embeddings