Earticle

초록

영어

Computer vision has rapidly evolved into a critical field that has garnered significant attention due to its applications in face recognition, human body analysis, automatic driving, indoor positioning, and other domains. The accuracy and speed of object detection have become a primary focus in computer vision research. Among the notable architectures, YOLO stands out, as it delivers remarkable speed that is 300 times faster than Fast-RCNN while maintaining comparable accuracy. In this paper, we proposed the topic of spatial recognition using the YOLO architecture. Specifically, we propose a solution that utilizes indoor video footage to identify objects in space, extract their spatial information, and store them in a database for matching and identifying spaces. We also introduce a new fingerprint input method that leverages monocular vision and YOLO algorithm to assist users in determining their location and space. Our study provides valuable insights and directions for future spatial recognition research.

목차

Abstract
1. Introduction
2. Related works
3. Yolov7-based approach
3.1. Object detection
3.2. Noise removal and data sorting
3.3 Build a digital map
3.4 Spatial recognition
4. Experiment setup
5. Experiment result
6. Conclusions
Acknowledgement
References

저자

• Haichuan Chen [ Dept. of AI Convergence Network, Ajou University ]
• Gaoyang Shan [ Dept. of Software and Computer Engineering, Ajou University ]
• Byeong-hee Roh [ Dept. of AI Convergence Network, Ajou University ] Correspondence author

참고문헌