This repository contains the code related to nf2vec framework, which is detailed in the paper Deep Learning on Object-centric 3D Neural Fields. In particular, here you can find the code regarding processing NeRFs. If you want to use the previous version of this framework for processing shapes, refer to inr2vec.

Before running the code, ensure that your machine is properly configured. This project was developed with the following main dependencies:

• python==3.8.18
• torch==1.12.0+cu113
• torchvision==0.13.0+cu113
• nerfacc==0.3.5 (with the proper CUDA version set)
• wandb==0.16.0

What follows are commands that you can execute to replicate the environment in which nf2vec was originally trained:

1. Install Python 3.8.18:

   conda install python=3.8.18

2. Install pip:

   conda install -c anaconda pip

3. Install PyTorch and torchvision:

   pip install torch==1.12.0+cu113 torchvision==0.13.0+cu113 --extra-index-url https://download.pytorch.org/whl/cu113

4. Install CUDA Toolkit:

   conda install -c "nvidia/label/cuda-11.7.1" cuda-toolkit

5. Install Ninja and Tiny CUDA NN:

   pip install ninja git+https://github.com/NVlabs/tiny-cuda-nn/#subdirectory=bindings/torch

6. Install NerfAcc:

   pip install nerfacc==0.3.5 -f https://nerfacc-bucket.s3.us-west-2.amazonaws.com/whl/torch-1.12.0_cu113.html

7. Install Einops:

   conda install -c conda-forge einops

8. Install ImageIO:

   conda install -c conda-forge imageio

9. Install WanDB:

   pip install wandb==0.16.0

10. Install h5py:

    conda install -c anaconda h5py

11. Install TorchMetrics:

    pip install torchmetrics

The generation task is based on a Latent GAN model detailed at THIS link. Please, follow the instructions provided at that link to properly configure your environment.

The mapping network task requires the training of the inr2vec framework. Please, refer to THIS page to properly configure your environment.

This section contains the details required to run the code.

IMPORTANT NOTES:

1. each module cited below must be executed from the root of the project, and not within the corresponding packages. This will ensure that all the paths used can properly work.

2. the file settings.py contains all the paths (e.g., dataset location, model weights, etc...) and generic configurations that are used from each module explained below.

3. Some training and experiments, such as the training of the nf2vec framework and the classification task, use the wandb library. If you want to use it, then you need to change the following two variables: os.environ["WANDB_SILENT"] and os.environ["WANDB_MODE"], which are located at the beginning of the settings.py module.

To train nf2vec you need to have a dataset of trained NeRFs. The implemented code expects that there exist the following files:

• data/train.json
• data/validation.json
• data/test.json

These JSONs hold a list of file paths, with each path corresponding to a NeRF model that has been trained, and then used in a specific data split. In particular, each path corresponds to a folder, and each folder contains the following relevant files:

• the trained NeRF's weights
• the NeRF's occupancy grid
• JSON files with transform matrices and other paramters necessary to train NeRFs.

The name of the files contained in these folders should not be changed. Within the repository, you can find the JSON files used to originally train the framework.

Execute the following command to train nf2vec:

python nerf2vec/train_nerf2vec.py

If you have enabled wandb, then you should update its settings located in the config_wandb method, which is localed in the train_nerf2vec.py module.

Execute the following command to export the nerf2vec's embeddings:

python nerf2vec/export_embeddings.py

Note that these embeddings are necessary for other tasks, such as classification, retrieval and generation.

Execute the following command to perform the retrieval task:

python task_interp_and_retrieval/retrieval.py

The results will be shown in the task_interp_and_retrieval/retrieval_plots_X folder, where X depends on the chosen split (i.e., train, validation or test). The split can be set in the main method of the retrieval.py module.

Each file created during a specific retrieval iteration will be named using the same prefix represented by a randomly generated UUID.

Execute the following command to perform the interpolation task:

python task_interp_and_retrieval/interp.py

The results will be shown in the task_interp_and_retrieval/interp_plots_X folder, where X depends on the chosen split (i.e., train, validation or test). The split can be set in the main method of the retrieval.py module.

Execute the following command to perform the classification task:

python task_classification/train_classifier.py

If you have enabled wandb, then you should update its settings located in the config_wandb method, which is localed in the train_classifier.py module.

In order to generate and visualize the new embeddings, it is necessary to execute some operations following a specific order.

The following command creates the folder task_generation/latent_embeddings, which will contain the nerf2vec's embedding properly organized for this task.

python task_generation/export_embeddings.py

The following command creates the folder task_generation/experiments, which will contain both the weights of the trained models and the generated embeddings:

python task_generation/train_latent_gan.py

All the hyperparameters used to train the Latent GANs can be found inside the train_latent_gan.py module.

The following command creates renderings from the embeddings generated during the previous step:

python task_generation/viz_nerf.py 

The renderings will be created in the GAN_plots_X folder, where X is the ID of a specific class.

Please refer to THIS README for this task.

Please contact us if you need access to the datasets, exported embeddings, and weights of the trained models used in all experiments.