Authors:
Nikolajs Bumanis
1
;
Gatis Vitols
2
;
Irina Arhipova
2
and
Egons Solmanis
2
Affiliations:
1
Faculty of Information Technologies, Latvia University of Life Sciences and Technologies, 2 Liela str., Jelgava, Latvia
;
2
WeAreDots Ltd., Elizabetes str. 75, Riga, Latvia
Keyword(s):
Multi-object Tracking, MOT Algorithms, Traffic Detection.
Abstract:
Visual object detection and tracking is a fundamental research topic in computer vision with multiple applications ranging from object classification to multi-object tracking in heavy urban traffic scenarios. While object detection and tracking tasks, especially multi-object tracking, have multiple solutions, it is still unclear how to build the real-world applications using different building blocks like algorithms, filters, base neuron networks, etc. The issue becomes more sophisticated as most of the recently proposed solutions are based on existing methodologies, frameworks and applicable technologies; however, some are showing promising results using contradictory realization. This paper addresses issues and research trends of multi-object tracking, while depicting its building blocks and currently best solutions. In result, a potential building blocks for real-world application in the framework of Jelgava city in Latvia is presented.