Open In Github

Python scripts for performing stereo depth estimation using the MobileStereoNet model in ONNX

Stereo depth estimation on the cones images from the Middlebury dataset (https://vision.middlebury.edu/stereo/data/scenes2003/)

Requirements

• Check the requirements.txt file. Additionally, pafy and youtube-dl are required for youtube video inference.
• DrivingStereo dataset, ONLY for the driving_sereo_test.pyscript. Link: https://drivingstereo-dataset.github.io/

Installation

pip install -r requirements.txt
pip install pafy youtube-dl

ONNX model

Download the ONNX model from Google Drive and save it into the models folder.

Original Pytorch model

The Pytorch pretrained model was taken from the original repository.

Examples

• Image inference:

 python image_depth_estimation.py 

• Video inference:

 python video_depth_estimation.py

• DrivingStereo dataset inference:

 python driving_sereo_test.py

Inference video Example

References:

• MobileStereoNet model: https://github.com/cogsys-tuebingen/mobilestereonet
• PINTO0309’s model zoo: https://github.com/PINTO0309/PINTO_model_zoo
• PINTO0309’s model conversion tool: https://github.com/PINTO0309/openvino2tensorflow
• DrivingStereo dataset: https://drivingstereo-dataset.github.io/
• Original paper: https://arxiv.org/pdf/2108.09770.pdf