The study introduces the CISTA-Flow network, integrating flow estimation with CISTA-LSTC for motion compensation in event-based video reconstruction. The iterative training framework enhances reconstruction accuracy and dense flow estimation simultaneously. Results show state-of-the-art performance over other advanced networks, with sharper edges and finer details in reconstructed frames. Different warped inputs improve temporal consistency, highlighting the importance of incorporating optical flow for motion compensation. The adaptability of CISTA-Flow to different flow networks demonstrates its potential for further improvements.
To Another Language
from source content
arxiv.org
Thông tin chi tiết chính được chắt lọc từ
by Siying Liu,P... lúc arxiv.org 03-19-2024
https://arxiv.org/pdf/2403.11961.pdfYêu cầu sâu hơn