- Given an audio description of the orientation of an object, the code should generate and renders the 3D model of that description.
- We used Spacy for NER recognition followed by Google's pre-trained word2vec embeddings to recommend the relevant entities to our target entity recognized from the scene. You can download Google’s pre-trained model here. It’s 1.5GB!
- K3D-Jupyter - a Jupyter notebook 3d visualization package used to render 3d objects.
Note: This repository code isn't optimized in any aspect.
- Audio Instructions to place 3d objects in the scene
- Example input:
Add me a table to the scene
- 3D models Rendering for Virtual Reality
- Suggest relevant items to the rendered scene that completes/creates a scene
