Document Type

Dataset

Rights

Available under a Creative Commons Attribution Non-Commercial Share Alike 4.0 International Licence

Grant Number

18/CRT/6222 - 713654

Disciplines

Computer Sciences

Publication Details

International Journal of Computer and Systems Engineering

https://github.com/fatmaelther/SatelliteDatasetsforRoadIntersectionDetection

Abstract

Understanding road networks plays an important role in navigation applications such as self-driving vehicles and route planning for individual journeys. Intersections of roads are essential components of road networks. Understanding the features of an intersection, from a simple T-junction to larger multi-road junctions is critical to decisions such as crossing roads or selecting safest routes. The identification and profiling of intersections from satellite images is a challenging task. While deep learning approaches offer state-of-the-art in image classification and detection, the availability of training datasets is a bottleneck in this approach. In this paper, a labelled satellite image dataset for the intersection recognition problem is presented. It consists of 14,692 satellite images of Washington DC, USA. To support other users of the dataset, an automated download and labelling script is provided for dataset replication. The challenges of construction and fine-grained feature labelling of a satellite image dataset is examined, including the issue of how to address features that are spread across multiple images. Finally, the accuracy of detection of intersections in satellite images is evaluated.

DOI

https://doi.org/10.21427/vtgv-5a40

Methodology

A constant problem in our work is lack of shared datasets. It is desirable to make the two datasets constructed here available to the research community. The terms of Google API prevent the direct publishing or distribution to others. In order for others researchers to access the dataset, we have provided an automated script to generate replica datasets via the Google maps static API, but using the API key of the downloader. The script will automatically download the exact same images for either/both satellite and hybrid – and apply the correct annotation of our dataset. The images used in these datasets were downloaded without cost, using the free allocation of images for Google maps static API, as at publication date. After capturing and annotating the dataset, two types of datasets were constructed: a satellite intersection dataset and a hybrid intersection dataset. Each one has two classes: intersection and

no-intersection. In each dataset, the intersection category contains 7342 images and the no- intersection category contains 7350 images.

This folder contains the annotation folder for our dataset and a script file:

  1. The annotation folder contains four files. Two of them are annotation files for testing images: one for intersection class and another one for no_intersection class. Other two of them are annotation files for training and validation images: one for intersection class and another one for no_intersection class.
  2. The script file is a python script to download images form google using maps static API and create dataset folder. After running this code, a dataset folder will be created.

To run this script, you need to enter the following parameters: api_key, maptype, annottation_folder_path, and destination_path.

  1. api_key: To download images form Google map, you need to create a project with a billing account. Then, you must enable the maps static API. Then, use this API key in our script. For more details about how to create this API, follow instructions in this link. https://developers.google.com/maps/documentation/maps-static/get-api-key
  2. maptype: you have to identify which type of image do you want to include in dataset. We used two type satellite or hybrid. Before you run the script, be sure that you use the one you want.
  3. annottation_folder_path: In this field, put the path for the annotation folder that associated with this script.
  4. destination_path: In this filed, define a path to a create dataset folder in it.

Language

English

File Format

csv,txt,ipynb

Viewing Instructions

To run this script, you need to enter the following parameters: api_key, maptype, annottation_folder_path, and destination_path. api_key: To download images form Google map, you need to create a project with a billing account. Then, you must enable the maps static API. Then, use this API key in our script. For more details about how to create this API, follow instructions in this link. https://developers.google.com/maps/documentation/maps-static/get-api-key maptype: you have to identify which type of image do you want to include in dataset. We used two type satellite or hybrid. Before you run the script, be sure that you use the one you want. annottation_folder_path: In this field, put the path for the annotation folder that associated with this script. destination_path: In this filed, define a path to a create dataset folder in it.

Data Owner

yes

Funder

Science Foundation Ireland,Marie Skłodowska-Curie

Share

 
COinS