Lightweight Optical Flow Estimation Using 1D Matching

doi:10.1109/ACCESS.2024.3380917

2024-01-01

Wonyong Seo

W.-T. Park

Munchurl Kim *

DOI：10.1109/ACCESS.2024.3380917

原文链接

原文求助

摘要

En 中文

Recent advancements in optical flow estimation have led to notable performance gains,driven by the adoption of transformer architectures, enhanced data augmentation, self-supervised learningtechniques, the use of multiple video frames and iterative refinement of estimate optical flows. Nonetheless,these cutting-edge methods encounter substantial challenges with surge in computational complexity andmemory demands. In response, we introduce a lightweight optical flow method, called MaxFlow, to addressthe trade-off between computational complexity and prediction performance. By leveraging MaxViT,we design a network with a global receptive field at reduced complexity, and proposed 1D matching toalleviate the computational complexity from (HxW)2toHxW(H+W), wherHandWdenotesheight and width of input image. Consequently, our method achieves the lowest computational complexitycompared to both state of the arts(SOTA) and other lightweight optical flow estimation methods, whilestill achieving competitive results with the SOTA techniques. We performed extensive experiments to showthe effectiveness of our method, achieving about 5 to 6 times reductions in computation complexity whilemaintaining the prediction accuracy with only degradation of 16% in term of end point error(EPE) at Sinteltest clean sequences with respect to RAFT method.

Keyword:

1D matching

MaxViT

optical flow estimation

lightweight optical flow estimation

positional embedding

transformer

期刊

IEEE Access

IF:

3.6

论文数:

9.8W

被引数:

29.4W

机构

samsung

学者数:

8.6K

论文数: 6.4K

被引数: 8

引用论文

Judgment Capacity, Fear of Falling, and the Risk of Falls in Community-Dwelling Older Adults: The Progetto Veneto Anziani Longitudinal Study 社区居住的老年人的判断能力，对跌倒的恐惧和跌倒的风险: Progetto Veneto Anziani纵向研究

Rejuvenation Research

IF0

2020-06-01

OAAI

Caterina Trevisan; Bruno M. Zanforlini; Stefania Maggi; Marianna Noale; Federica Limongi; Marina De Rui; Maria Chiara Corti; Egle Perissinotto; Anna-Karin Welmer; Enzo Manzato; Giuseppe Sergi

Effective Unconstrained Face Recognition by Combining Multiple Descriptors and Learned Background Statistics

IEEE Transactions on Pattern Analysis and Machine Intelligence

IF0

2011-10-01

PREAI

L. Wolf; T. Hassner; Y. Taigman

Enhanced CO₂/N₂ separation by supported ionic liquid membranes (SILMs) based on PDMS and 1-ethyl-3-methylimidazolium acetate

Chemical Engineering Communications

IF0

2019-11-22

PREAI

Z. Ziobrowski; A. Rotkegel