2025 LPCVC Overview
The 2025 IEEE Low-Power Computer Vision Challenge (LPCVC) marked a decade of advancing energy-efficient computer vision technologies. In collaboration with Qualcomm, this year’s competition featured three distinct tracks:
- Track 1: Image Classification under Various Lighting Conditions and Formats
- Track 2: Open-Vocabulary Segmentation with Text Prompt
- Track 3: Monocular Depth Estimation
The challenge emphasized practical applications on edge devices, encouraging participants to develop models that balance accuracy with low energy consumption. The submission window was open from March 1, 2025, to March 31, 2025. Winners were recognized for their innovative approaches and presented their solutions during the CVPR 2025 workshop.
A key feature of this year’s competition was the collaboration with Qualcomm, utilizing the Qualcomm AI Hub for model deployment. This partnership provided participants access to Qualcomm’s advanced AI tools and platforms, enabling efficient model development and deployment on edge devices.
2025 LPCVC Winners
The 2025 IEEE Low-Power Computer Vision Challenge (LPCVC) concluded successfully on 2025/03/31, after a month-long competition. A total of 59 teams, from 14 different countries and regions, submitted 516 solutions across three tracks, out of which 218 solutions outperformed the baseline solutions.
Winners for each track are as follows:
Track 1: Image Classification under Various Lighting Conditions and Formats
LabLVM – Score: 0.97406639, Time: 1.612 ms
Team members: Seungmin Oh, Hankyul Kang, Seunghun Kang, Jongbin Ryu
Track 2: Open-Vocabulary Segmentation with Text Prompt
SICer – Score: 0.5321869337, Time: 515.8 ms
Team members: Yuning Ji, Zizhou Tong, Zhuohang Li, Xinxin Wang, Chaoyao Shen, Linghui Kong, Chenlong Xia, Bohan Guo, Meng Zhang
Track 3: Monocular Depth Estimation
Sailor Moon – Score: 83.1397434, Time: 30.4 ms
Team members: Kexin Chen, Yuan Qi
Congratulations to all winners for their outstanding achievements in advancing low-power computer vision research.
8th Workshop on Efficient Deep Learning for Computer Vision
Sponsors
Organizers
| Name | Role and Organization | |
|---|---|---|
| Yung-Hsiang Lu | Professor, Purdue University | yunglu@purdue.edu |
| Zihao Ye | Purdue University | ye277@purdue.edu |
| Vincent Zhao | Purdue University | zhao1322@purdue.edu |
| George Thiruvathukal | Professor, Loyola University Chicago | gthiruvathukal@luc.edu |
| Mooi Choo Chuah | Professor, Lehigh University | mcc7@lehigh.edu |
| Bo Lang | Lehigh University | bol221@lehigh.edu |
| Zhen Yao | Lehigh University | zhy321@lehigh.edu |
| Zhihao Zheng | Lehigh University | zhzc21@lehigh.edu |
| Shuai Zhang | Qualcomm | shuazhan@qti.qualcomm.com |
| Xiao Hu | Qualcomm | hux@qti.qualcomm.com |
| Taotao Jing | Qualcomm | tjing@qti.qualcomm.com |
| Xin Li | Qualcomm | lxi@qti.qualcomm.com |
| Kory Watson | Qualcomm | kwatson@qti.qualcomm.com |
| Ashwin Murthy | Qualcomm | ashwmurt@qti.qualcomm.com |