There are a lot of technology for crowd sensing. Such as ultrasonic and pyroelectric infrared. These techniques have some shortcomings that they could not use on a large scale. Our purpose is to develop a sensor base carbon dioxide sensor group technology. Technology is used to evaluate the number of public transport and flow. The first is to set a carbon dioxide sensor and data collection. Then through the extreme learning machine (ELM) and Liner regression algorithm analysis data. We try to combine existing bus real-time information platform, refer to the number of real-time bus function. According to the traffic movements to improve bus stops and routes. To help with city planning.