https://github.com/oarriaga/bayesian-inverse-graphics

Bayesian Inverse Graphics for Few-Shot Concept Learning
https://github.com/oarriaga/bayesian-inverse-graphics

Last synced: about 1 month ago
JSON representation

Bayesian Inverse Graphics for Few-Shot Concept Learning

• Host: GitHub
• URL: https://github.com/oarriaga/bayesian-inverse-graphics
• Owner: oarriaga
• License: mit
• Created: 2024-06-20T13:30:25.000Z (5 months ago)
• Default Branch: main
• Last Pushed: 2024-08-01T09:21:14.000Z (4 months ago)
• Last Synced: 2024-08-01T10:50:40.062Z (4 months ago)
• Language: Python
• Size: 52.7 KB
• Stars: 6
• Watchers: 2
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        
# Bayesian Inverse Graphics (BIG)

This repository contains the code for the paper "Bayesian Inverse Graphics for Few-Shot Concept Learning"

TLDR: `probabilistic programming` + `differentiable rendering` = `minimal-data learning`

## Modules 

All modules are implemented in ```jax```

* [jaynes](https://github.com/oarriaga/bayesian-inverse-graphics/tree/main/jaynes) Probabilistic Programming Library (Automatic Bayesian Inference).

* [tamayo](https://github.com/oarriaga/bayesian-inverse-graphics/tree/main/tamayo) Differentiable Rendering Library.

* [lecun](https://github.com/oarriaga/bayesian-inverse-graphics/tree/main/lecun) Convnets.

## Run

### Setup

0. Install requirements e.g. `pip install -r requirements.txt` 

1. Download the datasets (fscvlr.zip) and weights (VGG16.eqx) from [here](https://github.com/oarriaga/bayesian-inverse-graphics/releases/tag/v0.0.1).

2. Move `fsclvr.zip` inside repository `bayesian-inverse-graphics/`.  

3. Move `VGG16.eqx`  inside repository `bayesian-inverse-graphics/`.  

4. Extract datasets `unzip fsclvr.zip`

### Training

5. Run `python optimize_scene.py`

5. Run `python extract_features.py`

7. Run `python optimize_bijectors.py`

### Test

8. Run `python learn_concept.py --concept 0`

## Funding

This project was developed in the [Robotics Group](https://robotik.dfki-bremen.de/de/ueber-uns/universitaet-bremen-arbeitsgruppe-robotik.html) of the [University of Bremen](https://www.uni-bremen.de/), together with the [Robotics Innovation Center](https://robotik.dfki-bremen.de/en/startpage.html) of the **German Research Center for Artificial Intelligence** (DFKI) in **Bremen**.

It has been funded by the German Federal Ministry for Economic Affairs and Energy and the [German Aerospace Center](https://www.dlr.de/DE/Home/home_node.html) (DLR), in the [PhysWM](https://robotik.dfki-bremen.de/en/research/projects/physwm) project.