Turan, ErhanDandil, BesirAvci, Engin2024-09-182024-09-182022978-3-030-94191-8978-3-030-94190-12367-33702367-3389https://doi.org/10.1007/978-3-030-94191-8_17https://hdl.handle.net/20.500.12483/126216th International Conference on Smart City Applications -- OCT 27-29, 2021 -- Safranbolu, TURKEYThe main reason for congestion in traffic is unnecessary waiting time at intersections. Economic and environmental improvement can be directly achieved by reducing the waiting time at the junction points. The controller, in which the signaling values are calculated by taking the vehicles and pedestrians into consideration in social terms, needs to be developed. Maximum flow in traffic at a single junction can cause a bottleneck at the next junction. Therefore, junction signaling times should be calculated in relation to each other for a certain region and line. Rule and learning based methods cannot respond to multiple intersections and unlikely situations. In order to provide maximum flow dynamically, a new Graph algorithm based on deep learning is needed. In this study, a deep learning-based Graph method has been proposed in which rule and learning-based methods will be used against unlikely situations and to eliminate the single junction bottleneck disadvantage.eninfo:eu-repo/semantics/closedAccessTraffic junction signalingDeep learningGraph algorithmMax flow algorithmConference Object39321122110.1007/978-3-030-94191-8_172-s2.0-85126342357Q4WOS:000928840400017N/A