Learning depth from video in dynamic scenes

Paper Title:

Learning Monocular Depth in Dynamic Environment via Context-aware Temporal Attention

Published on:

12 May 2023

Primary Category:

Computer Vision and Pattern Recognition

Paper Authors:

Zizhang Wu,

Zhuozheng Li,

Zhi-Gang Fan,

Yunzhe Wu,

Yuanzhu Gan,

Jian Pu,

Xianzhi Li

Bullets

Key Details

•

Proposes context-aware temporal attention to handle moving objects in video

•

Uses long-range geometry embedding for improved reasoning

•

Achieves state-of-the-art accuracy on depth estimation benchmarks

•

Focuses on robustness in dynamic scenes with moving objects

•

Employs attention mechanisms for feature integration

Explore the topics in this paper

deep learning

depth estimation

dynamic scenes

geometry

monocular video

AI generated summary

Learning depth from video in dynamic scenes

This paper proposes a deep learning method to estimate depth from monocular video frames. It handles dynamic scenes with moving objects by using context-aware temporal attention and long-range geometry embedding. The method achieves state-of-the-art depth estimation accuracy.