Open AccessComputer Vision: A Review of Detecting Objects in Videos – Challenges and Techniques
Author(s) -
Mohammad Ali A. Hammoudeh,
Mohammad Alsaykhan,
Ryan Alsalameh,
Nahs Althwaibi
Publication year - 2022
Publication title -
international journal of online and biomedical engineering
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.184
H-Index - 8
ISSN - 2626-8493
DOI - 10.3991/ijoe.v18i01.27577
Subject(s) - computer science , scope (computer science) , pedestrian , domain (mathematical analysis) , artificial intelligence , object (grammar) , set (abstract data type) , computer vision , video tracking , tracking (education) , human–computer interaction , object detection , computer security , intelligent transportation system , transport engineering , engineering , pattern recognition (psychology) , psychology , mathematical analysis , pedagogy , mathematics , programming language
Traffic safety aims to change the attitude of citizens towards careless traffic on the roads, making this the first step towards changing behavior. Also, teach the rules of safe pedestrian behavior and minimize the risks of road accidents. So many regulations have been set to avoid road accidents and traffic jams, which is the study scope of this paper using IT technology. With the expanding interests in Computer vision use cases such as vehicles self-driving, face recognition, intelligent transportation frameworks and so on individuals are hoping to assemble custom AI models to recognize and distinguish specific objects. Object detection is part of a computer's vision where objects that can be observed externally and are found in videos can be identified and tracked by computers. Therefore, object tracking is an important part of video analysis. There are many proposed methods such as Tracking, Learning, Detection, Mean shift and MIL. In this paper, the computer vision state in object detecting domain along with its challenges are discussed, also we address some requirements and techniques to overcome these challenges. Finally, TensorFlow technology is presented as a recommended solution to support Lane’s violation.