Project DetailsThis work deals with the identification of pedestrians approaching a crosswalk when they use a mobile phone. As part of the research work of Michal Derhy from the Faculty of Architecture and Urban Planning, the option of creating a system that alerts distracted people approaching to cross the road was examined.
In an experiment conducted by Michal, a system was set up that included an alarm that was activated manually when a person that use the phone approached the crosswalk.
The results of the experiment showed that there is a direct effect between the warning to the pedestrians and their alertness to what is happening around the crosswalk. Based on the results, it was decided to examine the possibility of creating an automatic system that can detect and alert when a distracted pedestrian approaches a crosswalk.
Therefore, the following project goal was defined: proof of concept for an automatic system that can identify pedestrians who are distracted from using a mobile phone, when they approach the crosswalk.
The system is based on the location of a stationary camera near a crosswalk, the photo goes to a processing system based on a deep network which analyses the video, and finally gives an indication if the person approaching the crosswalk is using a mobile phone.
The network chosen for use in the project was required to classify people's actions with an emphasis on phone use, and in particular the following actions were examined: looking at phone, talking on cell phone, texting. Based on these requirements and the examination of the performance, the Temporal Segment Networks (TSN) network was chosen for the classification. As part of the solution to the problem, a part of pre-processing and a part of post-processing were added in order to adapt to the requirements of the project.
The system was tested on a dedicated data base collected by us in which people were filmed approaching a crosswalk while using the phone, and the results showed that we can give a correct indication with a success rate of about 80%.
