Yifei Liu

RI-SQH, CMU.

1723 Murray Ave

Pittsburgh, PA 15217

Hello. I’m a Master’s student in Robotics at Carnegie Mellon University, advised by Prof. Sebastian Scherer in the AirLab. I am currently working on 3D perception for UAV and UGV in visually-degraded, outdoor environments. I am particularly interested in the intersection of computer vision, deep learning, and robotics.

After my first year of graduate studies, I interned at NASA Jet Propulsion Laboratory, working on perception for offroad navigation. I received my Bachelor’s degree in Computer Science from the University of California, Irvine, where I worked with Prof. Roy Fox in the Intelligent Dynamics Lab on semantic perception and sim2Real for vision-based reinforcement learning. During my undergrad, I also developed computer vision applications at the UCI Center for Artificial Intelligence in Diagnostic Medicine and USC Energy Efficient Secure Sustainable Computing Group.

Check out my research and projects here.

News

Sep 12, 2024	FIReStereo is live!
May 13, 2024	Started interning at NASA Jet Propulsion Laboratory, Robotics Systems Section
Apr 23, 2024	Sponsored by NASA I-Corps to attended the IDGA Wildfire Technology Summit
Aug 21, 2023	Joined Carnegie Mellon University for M.S. in Robotics
Mar 24, 2023	Graduated from University of California, Irvine with distinction

Publications

FireFly: A Synthetic Dataset for Ember Detection in Wildfire

Yue Hu, Xinan Ye, Yifei Liu, and 7 more authors

IEEE/CVF International Conference on Computer Vision (CVPR), 2023

Abs arXiv PDF

This paper presents "FireFly", a synthetic dataset for ember detection created using Unreal Engine 4 (UE4), designed to overcome the current lack of ember-specific training resources. To create the dataset, we present a tool that allows the automated generation of the synthetic labeled dataset with adjustable parameters, enabling data diversity from various environmental conditions, making the dataset both diverse and customizable based on user requirements. We generated a total of 19,273 frames that have been used to evaluate FireFly on four popular object detection models. Further to minimize human intervention, we leveraged a trained model to create a semi-automatic labeling process for real-life ember frames. Moreover, we demonstrated an up to 8.57% improvement in mean Average Precision (mAP) in real-world wildfire scenarios compared to models trained exclusively on a small real dataset.
FIReStereo: Forest InfraRed Stereo Dataset for UAS Depth Perception in Visually Degraded Environments

Devansh Dhrafani^*, Yifei Liu^*, Andrew Jong, and 6 more authors

ArXiv, 2024

Abs arXiv Website

Robust depth perception in degraded visual environments is crucial for autonomous aerial systems. Thermal imaging cameras, which capture infrared radiation, are robust to visual degradation. However, due to lack of a large-scale dataset, the use of thermal cameras for unmanned aerial system (UAS) depth perception has remained largely unexplored. This paper presents a stereo thermal depth perception dataset for autonomous aerial perception applications. The dataset consists of stereo thermal images, LiDAR, IMU and ground truth depth maps captured in urban and forest settings under diverse conditions like day, night, rain, and smoke. We benchmark representative stereo depth estimation algorithms, offering insights into their performance in degraded conditions. Models trained on our dataset generalize well to unseen smoky conditions, highlighting the robustness of stereo thermal imaging for depth perception. We aim for this work to enhance robotic perception in disaster scenarios, allowing for exploration and operations in previously unreachable areas.