I am a Master (Sep. 2021 - Mar. 2024) student at College of Control Science and Engineering, Zhejiang Unerveristy.
Before that, I obtatin my B.Eng (Sep. 2017 - Jun. 2021) from College of Instrumentation and Electrical Engineering, Jilin Unerveristy.
My research interest lies in 3D Vision, especially in the field of Neural Rendering, Implicit Representation, and SLAM.
I also love 3D/4D Object Generation! Please check my github and see my implementation for Diffusion Model.
Representative papers are highlighted.
Humans can infer 3D structure from 2D images of an object based on past experience and improve their 3D understanding as they see more images.
Inspired by this behavior, we introduce SAP3D, a system for 3D reconstruction and novel view synthesis from an arbitrary number of unposed images.
Adaptive Positional Encoding is proposed in this paper to train neural radiance fields from unknwon camera poses (or even initrinics).
The theoretical relationship between Positional Encoding and Fourier Series Regression is investigated to prove that learnable frequencies can improve both accuracy of camera parameter estimation and view synthesis quality of scene representation.
In this paper, we exploit the geometric, semantic, and photometric guidance to represent the neural radiance fields from sparse view inputs.
We propose hierarchical geometric guidance (HGG) to sample volume points with the depth prior and hierarchical semantic guidance (HSG) to supervise semantic consistency of the complex real-world scenarios using CLIP.
Thermal images are used in this paper to realize a robust visual SLAM system in challenging environment.
The proposed method, a thermal-inertial SLAM system, represents several improvements, including SVD-based image processing and Thermal-RAFT tracking methods.