인문학
사회과학
자연과학
공학
의약학
농수해양학
예술체육학
복합학
DBpia AI
AI 에이전트
AI 아이디어
AI 리더
AI 뷰어
Citeasy
크롬 서지관리 다운로드
지원사업
학술연구/단체지원/교육 등 연구자 활동을 지속하도록 DBpia가 지원하고 있어요.
커뮤니티
연구자들이 자신의 연구와 전문성을 널리 알리고, 새로운 협력의 기회를 만들 수 있는 네트워킹 공간이에요.
논문 기본 정보
- 자료유형
- 학술대회자료
- 저자정보
- 발행연도
- 2024.9
- 수록면
- 193 - 197 (5page)
이용수
초록· 키워드
Gaze estimation plays a critical role in various applications, including virtual reality, human-computer interaction, and advanced driver monitoring systems. This paper introduces a Gaze Estimation Model that employs a dual-stream architecture, integrating RGB and depth data to achieve precise 3D gaze prediction. The model leverages EfficientNet-B3 as the backbone on both streams and incorporates Long Short-Term Memory (LSTM) layers to capture temporal dependencies, effectively enhancing performance in dynamic environments. Through comprehensive evaluation of the EyeDiap dataset, the proposed model achieves a Mean Angular Error (MAE) of 5.96°, outperforming existing state-of-the-art methods and setting a new benchmark for 3D gaze estimation accuracy. The integration of depth data with RGB imagery enriches the feature set, significantly contributing to the model’s accuracy. Despite challenges such as increased computational demands and noise in-depth data, the model demonstrates robust, real-time performance, making it suitable for deployment in real-world applications. Future work will aim to optimize computational efficiency and extend the model’s applicability across diverse conditions.
#Gaze Estimation
#Dual-Stream Architecture
#RGB-D Imaging
#EfficientNet-B3
#EyeDiap Dataset
#Real-Time Prediction
#Multi-Modal Deep Learning
상세정보 수정요청해당 페이지 내 제목·저자·목차·페이지정보가 잘못된 경우 알려주세요!
목차
- Abstract
- I. INTRODUCTION
- II. RELATED WORK
- III. METHODOLOGY
- IV. EXPERIMENTS
- V. CONCLUSION
- REFERENCES
참고문헌
참고문헌 신청최근 본 자료
UCI(KEPA) : I410-151-25-02-090961773
