Comprehending object detection by deep learning methods and algorithms.

Priyanka, Mallineni; Lavanya, Kotapati; Sai, K. Charan; Rohit, Kavuri; Bano, Shahana

Abstract

In the real world, computer vision is used for more challenging tasks like detecting of objects in an image or video. There are multiple applications of object detection in various domains like animation, autonomous driving, monitoring of traffic, communicating through video. With the development of new emerging technologies in deep learning, finding accuracy of objects by performing classification and detection became possible. When compared to traditional object detection methods deep learning methods has an ability of feature learning and rendering. This paper is mainly focused on the working procedure of convolutional neural networks in detecting objects that are present in the environment of an image. CNN, R-CNN, and Faster R-CNN are the main models of deep learning which are considered for comparative-based study. Comparison between these models is made by identifying their accuracies, limitations, and speed. Among the three models, Faster R-CNN is identified as ideal one as it has higher accuracy and less expensive in nature when compared with R-CNN whereas CNN model can be only used for image classification (Tripathi in Journal of Innovative Image Processing (JIIP) 3:100–117, 2021), but it cannot localize the objects.

Citation

PRIYANKA, M., LAVANYA, K., SAI, K.C., ROHIT, K. and BANO, S. 2022. Comprehending object detection by deep learning methods and algorithms. In Shakya, S., Ntalianis, K. and Kamel, K.A. (eds.) Proceedings of the 3rd International conference on mobile computing and sustainable informatics (ICMCSI 2022), 27-28 January 2022, Kirtipur, Nepal. Lecture notes on data engineering and communications technologies, 126. Singapore: Springer [online], pages 523-537. Available from: https://doi.org/10.1007/978-981-19-2069-1_36

• Presentation Conference Type: Conference Paper (published)
• Conference Name: 3rd International conference on mobile computing and sustainable informatics (ICMCSI 2022)
• Start Date: Jan 27, 2022
• End Date: Jan 28, 2022
• Acceptance Date: Dec 15, 2021
• Online Publication Date: Jul 15, 2022
• Publication Date: Dec 31, 2022
• Deposit Date: Aug 13, 2024
• Publicly Available Date: Aug 13, 2024
• Publisher: Springer
• Peer Reviewed: Peer Reviewed
• Pages: 523-537
• Series Title: Lecture notes on data engineering and communications technologies
• Series Number: 126
• Series ISSN: 2367-4512 ; 2367-4520
• ISBN: 9789811920684
• DOI: https://doi.org/10.1007/978-981-19-2069-1_36
• Keywords: Deep learning; Convolutional neural networks; Region-based convoltuional neural networks; Selective search
• Public URL: https://rgu-repository.worktribe.com/output/2063933