Transtouch

Paper | Data | Video

This is the official repo for the implementation of Transtouch: Learning Transparent Object Depth Sensing Through Sparse Touches .

Usage

Data Convention

The data is organized as follows:

<sceneId_viewId>
|-- meta.pkl    # scene parameters
|-- 1024_irL/R_*.png # images taken from left/right camera with different exposure
|-- label*.png  # labels of objects
|-- 1024_irL/R_real_templcn2_ps11_t0.005.png # generated binary pattern

We use split files in ./split_files to select scenes and views for train and test.

Setup

Clone this repository

git clone https://github.com/ritsu-a/transtouch
cd transtouch
pip install -r requirements.txt

Running

Replace $MODEL_PATH with the path where weights are stored.

• Finetuning

python transtouch/tune.py --cfg configs/baseline.yml -s TEST.WEIGHT $MODEL_PATH

• Testing

python transtouch/test.py --cfg configs/baseline.yml -s TEST.WEIGHT $MODEL_PATH

Citation

Cite as below if you find this repository is helpful to your project:

Acknowledgement

Some code snippets are borrowed from Activezero2. The readme file is modified based on NeuS Thanks for these great projects.