Installing Requarements

foo@bar:~$ pip3 install -r requirements.txt

Installing kaolin lib:

(also here is documentation)

foo@bar:~$ pip install torch==2.1.1 torchvision==0.16.1 torchaudio==2.1.1 --index-url https://download.pytorch.org/whl/cu121
foo@bar:~$ pip install kaolin==0.15.0 -f https://nvidia-kaolin.s3.us-east-2.amazonaws.com/torch-2.1.1_cu121.html

PreCERMIT model

This model generates point cloud with 1024 points from given arbitrary number of photos/renders from different views.

directory with model with all scripts: https://github.com/bananananacat/Generation-of-3D-Objects/tree/main/model/models/PreCERMIT

CERMIT model

This model generates point cloud with N*k points from given point cloud with N points(quality enhansment).

directory with model with all scripts: https://github.com/bananananacat/Generation-of-3D-Objects/tree/main/model/models/CERMIT

Losses && Metrics

You can read all metrics in paper

Dataset

We created our own dataset and used ShapeNet dataset. There are some scripts - render functions, functions to work with camera and lights, functions to create dataset, etc.

Paper

Paper got 2 weak rejects and 1 strong reject on WACV 2025, the main drawback was that we didn't have enough time to complete some of the experiments and write them down in the paper :(

Despite this, the idea of ​​the architectures(PreCERMIT + CERMIT) was good