This is the software repository for the BusStopCV project. Please cite as:
Minchu Kulkarni, Chu Li, Jaye Ahn, Katrina Ma, Zhihan Zhang, Michael Saugstad, Yochai Eisenberg, Valerie Novack, Brent Chamberlain, Jon E. Froehlich. 2023. BusStopCV: A Real-time AI Assistant for Labeling Bus Stop Accessibility Features in Streetscape Imagery. In The 25th International ACM SIGACCESS Conference on Computers and Accessibility (ASSETS ’23), October 22–25, 2023, New York, NY, USA. ACM, New York, NY, USA, 5 pages
BusStopCV is an innovative solution that aims to simplify the process of labeling bus stop accessibility features. Our goal is to empower individuals without specialized expertise by offering a flexible crowdsourcing platform. This platform, combined with advanced AI technology, makes it easy to quickly and precisely label important bus stop details. These labeled datasets can seamlessly fit into modern mapping tools, improving navigation experiences for everyone.
BusStopCV primarily revolves around four key bus stop features: seating, shelter, signage, and trash cans. By combining crowdsourcing dynamics with advanced AI technologies, users receive guidance in their labeling efforts, ensuring accuracy and effectiveness. The collaboration between user input and AI-generated insights is improved through manual verification and correction, leading to detailed annotations.
Our thorough study highlights the exciting possibilities of the crowd+AI approach in labeling bus stop features. By utilizing real-time computer vision within web browsers, BusStopCV represents a significant step towards better awareness of accessibility. Supported by concrete results, this technology not only improves the precision of feature annotations but also speeds up the process, ultimately enhancing modern mapping tools with essential accessibility data.
BusStopCV is built in JavaScript (frontend) and Java (backend) with a custom-trained YOLOv8 model for computer vision.
1. Clone repository
Run git clone https://github.com/ProjectSidewalk/BusStopCV.git in the directory where you want to place the code.
2. Install Apache Tomcat
To run the application, Apache Tomcat must be installed on your computer. See the Tomcat webpage to select the correct version.
Follow these stpes to install Apache Tomcat using the terminal:
Download Apache Tomcat using the command wget https://downloads.apache.org/tomcat/tomcat-10/vVERSION/bin/apache-tomcat-VERSION.tar.gz replacing VERSION with your specified Apache Tomcat version
Extract the tarball with the command tar -xvf apache-tomcat-VERSION.tar.gz
Navigate to the Tomcat bin directory. Running ./startup.sh within this directory will start Tomcat.
Open a web browser and visit http://localhost:8080/. If Tomcat is correctly installed and started, the Tomcat homepage should be visible
To stop Tomcat run ./shutdown.sh within the bin directory of your Tomcat directory
3. Install maven
Install Apache Maven by running sudo apt install maven in the terminal
1. Start tomcat: Start Tomcat by navigating to the Tomcat bin directory and start Tomcat by running the startup.sh script
2. Create a clean maven installation of the project: Run the command mvn clean install within the BusStopCV directory to create a clean maven installation of the project
3. Copy .war file to webapps directory: Run the command cp target/accesslabeler-1.0-SNAPSHOT.war path_to_your_tomcat/webapps/accesslabeler.war to copy the generated .war file to your Tomcat webapps directory
4. Open application on browser: Visit http://localhost:8080/accesslabeler to access the BusStopCV web application.
5. Stop Tomcat: Once finished with the web application, shut down Tomcat by navigating to the Tomcat bin directory and running the shutdown.sh script