CHSwiss Federal Railways (SBB) are testing a system developed by the ETH spin-off Parquery to improve the accuracy of passenger load indicators in their mobile app. Currently, these indicators rely on predictive models based on historical data, which can sometimes result in inaccurate estimates of crowding levels. The new approach utilizes existing security cameras and artificial intelligence to count passengers boarding trains in real-time, claiming a precision rate of 98 percent. Parquery emphasizes that the technology does not perform facial recognition or store video data, operating strictly separately from security surveillance systems. While initial tests have revealed challenges such as tracking individuals who remove hats while boarding, the company states that the AI can be trained to overcome these issues. SBB describes the project as a proof of concept, noting that deployment could take until 2029 and remains uncertain, with plans to consult federal data protection authorities early in the process.
Bias read (Center): The article maintains a neutral tone by balancing technical claims with corporate caution and referencing past privacy controversies without taking a side.
