ViTaM Algorithm

This repository mainly contains the deep-learning algorithm in the paper:

Capturing forceful interaction with arbitrary objects using a deep learning-powered stretchable tactile array

Get-Started

The code is only tested on Ubuntu, we will soon test it on Windows system.

With conda and pip

Install anaconda or miniconda. Supposing that the name vtaco is used for conda environment:

conda create -y -n vitam python=3.9
conda activate vitam

Then, install dependencies with pip install

pip install -r requirements.txt

In order to use the manopth module, go to the directory and install it

cd components/manopth
pip install .

Dataset

We have prepared a small data sample in the directory data, with an object in the bottle category and a sequence of a hand grabing it. We will release the whole dataset sooner.

Training

To train on this small data sample, simply use the following command

python train.py --config-name test

It will create a directory outputs/test which will generate the output of the training process, including tensorboard log files, visualization results, etc.