โ๏ธโ๏ธ Article: "๐๐ ๐ฝ๐น๐ผ๐ฟ๐ถ๐ป๐ด ๐๐ฑ๐๐ฎ๐ป๐ฐ๐ฒ๐ฑ ๐๐ก๐ก ๐ฎ๐ป๐ฑ ๐ง๐ฟ๐ฎ๐ป๐๐ณ๐ผ๐ฟ๐บ๐ฒ๐ฟ-๐๐ฎ๐๐ฒ๐ฑ ๐๐ฟ๐ฐ๐ต๐ถ๐๐ฒ๐ฐ๐๐๐ฟ๐ฒ๐ ๐ณ๐ผ๐ฟ ๐ข๐ฟ๐ถ๐ฒ๐ป๐๐ฒ๐ฑ ๐ฆ๐บ๐ฎ๐น๐น ๐ข๐ฏ๐ท๐ฒ๐ฐ๐ ๐๐ฒ๐๐ฒ๐ฐ๐๐ถ๐ผ๐ป ๐ถ๐ป ๐๐ฒ๐ฟ๐ถ๐ฎ๐น ๐๐บ๐ฎ๐ด๐ฒ๐ฟ๐"
๐จโ๐ป๐ฉโ๐ป STUDENTS:
โข Nguyen Xuan Quang - KTPM2022 - Co-author
โข Le Toan - KTPM2022 - Co-author
โข Tran Nguyen Chi Huy - KTPM2022 - Co-author
โข Nguyen Vu Binh - KTPM2022 - Co-author
๐จโ๐ผ๐ฉโ๐ผ Supervisor:
โข Dr. Nguyen Tan Tran Minh Khang
Summary:
Research paper This paper evaluates the Oriented RepPoints method โ a technique for detecting small objects with arbitrary orientations in overhead images. The paper experiments with various network architectures such as ResNet, ConvNeXt, and PVT, in which Simple models like ResNet-50 and ConvNeXt show better results in detecting small and oriented objects. The paper also points out some challenges like objects with similar shapes and limited training data, which provides better insights into network architecture choices in real-world applications.
"We would like to sincerely thank the Faculty of Software Engineering, the Multimedia Communication Laboratory and the UIT-Together research group for creating conditions that helped us to research and complete this article."
- ----------------------------------
Journal of Computing and Information Technology is a Q4 ranked Journal at Scimago, Scopus category
