Pedestrian tracking and re-identification on top-view multi-camera system

• Autor(a) principal: Bóia, Rui Gonçalo Barreto Abrantes
• Data de Publicação: 2018
• Tipo de documento: Dissertação
• Idioma: eng
• Título da fonte: Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos)
• Texto Completo: http://hdl.handle.net/10773/25978
• topic: Computer vision; Re-identification; Pedestrian tracking

Understanding movement patterns inside retail areas is an information that is very useful nowadays, with several systems already developed to count and track pedestrians in those areas. The goal of this dissertation is to develop a system capable of counting, tracking, but mainly re-identifying pedestrians across multiple cameras with top-view, so that we can detect and analyze movement patterns in retail areas or similar applications. The work was divided in four important stages: the study and understanding of various systems that are already developed and the available algorithms; choosing the path to take in order to complete the proposed goals; video capture of adequate footage in order to simulate real scenarios as best as possible; test the developed algorithms and validate the obtained results. Results show that this system is capable of extracting interesting information regarding movement patterns and also capable to re-identify pedestrians across multiple cameras. However, this system might need to be adapted for real life scenarios with a bigger number of pedestrian passages.