Object Recognition with Machine Learning: Case Study of Demand-Responsive Service

To assist the disabled, a system was developed and designed to provide passenger transport services for disabled persons in wheelchairs. Machine-learning image-recognition technology was applied to provide accessible bus rides for the disabled. Based on the video footage at the bus stop, this system can judge whether there are wheelchair riders in the waiting area. YOLOv3, a real-time image-recognition model, was used for object recognition and notification. By combining LINE with the Flask Web application framework, it helps the disabled transmit messages and actively notifies bus drivers. The API server was established using the Flask framework. When the system at the bus stop detects the disabled, the images of the disabled waiting for the bus are transmitted to the server, and the images and related messages are transmitted in real time to the users through the LINE Message API. Passengers can also use the LINE chatbot to enter keywords to confirm the real-time location of the bus. The dynamic bus information comes from the Open Data of the Bureau of Transportation. The LINE BOT API program was developed using the PyCharm tool and then transmitted to the Heroku cloud platform for deployment. The LINE BOT API provides real-time wheelchair image recognition, and receives notifications and bus information queries through the LINE chatbot. The main contributions of this study are the use of artificial intelligence (AI) technology to provide solutions to facilitate bus rides for the disabled and to improve the quality of barrier-free transport services in modern society.