Towards Automatic Assessment of Perceived Walkability (AAPW)

This repository contains supplementary material for the paper:

Towards automatic assessment of perceived walkability

Ivan Blečić, Arnaldo Cecchini, Giuseppe A. Trunfio

Published in: International Conference on Computational Science and Its Applications (ICCSA 2018), Springer International Publishing, pp. 351-365

📄 Paper: Link to Springer

Abstract

We present a method for automatic assessment of perceived walkability by pedestrians, using a machine learning technique with deep convolutional neural networks (CNNs) trained on a dataset of georeferenced street-level images obtained from Google Street View.

Data

• streetview.zip - Dataset of georeferenced street-level images from Google Street View used for training, validation and testing of the CNN model.

Data Collection and Privacy

The data provided in this repository has been fully anonymized to ensure compliance with privacy regulations and ethical standards. All personally identifiable information has been removed or irreversibly transformed.

The human assessments used to build the training dataset were collected through anonymous online surveys conducted via a custom web-based interface. Participants voluntarily rated street-level images on a scale of perceived walkability, without providing any personal information. No tracking cookies, IP addresses, or other identifying data were collected during the survey process. The survey protocol was designed to ensure complete anonymity of respondents.

Citation

If you use this data or code, please cite:

@inproceedings{blecic2018towards,
  title={Towards automatic assessment of perceived walkability},
  author={Ble{\v{c}}i{\'c}, Ivan and Cecchini, Arnaldo and Trunfio, Giuseppe A},
  booktitle={International Conference on Computational Science and Its Applications},
  pages={351--365},
  year={2018},
  publisher={Springer}
}

License

Please refer to the paper for terms of use of the data.