RT info:eu-repo/semantics/masterThesis
T1 Análisis en tiempo real del flujo de pasajeros en las estaciones de tranvía mediante IA y visión por computador
A1 Oliva García, Jaime
A2 Máster Universitario en Ingeniería Industrial
AB The present work addresses the development of a person detection and tracking systemusing the Jetson Nano platform, Raspberry PI V2.1 camera and artificial intelligencemodels based on YOLOv8. The main objective of the project is to implement a systemcapable of detecting and counting people in various environments through the analysisof pre-recorded videos. To achieve this, two main components were developed: a mainscript and a tracking module.The main script is responsible for video capture, object detection, definition of areas ofinterest, and visualization of results. Several YOLOv8 models (from YOLOv8n to YOLOv8l)were employed, with YOLOv8s finally being selected for its optimal balance betweenaccuracy and speed. The tracking module, on the other hand, tracks the detected peopleacross video frames, assigning unique identifiers and maintaining continuous tracking.Tests were conducted with two pre-recorded videos: one of a plaza and another of atrain station. In the first case, the system successfully detected and counted people with100% accuracy. In the second case, although the detection was satisfactory, the countingwas not as accurate due to the lower video quality and greater recording distance.In conclusion, the results obtained demonstrate the effectiveness of the developedsystem and suggest future improvements, such as real-time implementation, model optimization,and the detection of other objects, which would expand the system’s capabilitiesfor security and flow monitoring applications.
YR 2024
FD 2024
LK http://riull.ull.es/xmlui/handle/915/39360
UL http://riull.ull.es/xmlui/handle/915/39360
LA es
DS Repositorio institucional de la Universidad de La Laguna
RD 24-nov-2024