2023 LPCVC Overview

For the last eight years, the IEEE Low Power Computer Vision Challenges has been attracting an increasing number of top researchers worldwide to show their solutions that can achieve high accuracy with low energy consumption. The 2023 Low-Power Computer Vision Challenge (2023 LPCVC) was hosted online and featured a single track designed by ACM Sigda and IEEE CEDA.

The task was semantic segmentation of UAV-view disaster scenes(a) into different regions associated with semantic categories related to disasters (b). The dataset consisted of 1,720 512x512 resolution images with 1,120 images released to participants as training and validation data and 600 images were kept private for testing. The models had to segment these images into a 512x512 prediction map. There were 14 different semantic categories. The model was tested on a Jetson Nano 2GB Developer Kit (power mode 5W).

The challenge was be open for submission from 2023/07/01-2023/08/04, as it was extended by four days due to a power outage on 07/10, which caused the server to crash. Winners will present their solutions during the Embedded Systems Week, which will be held between September 17-22, 2023. Please join the online discussion forum on the Slack.

*The participant agreement for the LPCVC 2023 contest can be found here.

2023 LPCVC Winners Announcement

The 2023 IEEE Low-Power Computer Vision Challenge (LPCVC) concluded successfully on 2023/08/04, after over a month of competition; 117 teams submitted 679 solutions to the competition, out of which 60 teams and 201 solutions outperformed the sample solution: On-Device Disaster Scene Parsing.

The winners of the On-Device Disaster Scene Parsing track are:

  1. ModelTC - Score: 75.608, Accuracy: 0.51162, Time: 6.8 ms.
    The team members are Yiru Wang, Xin Jin, Zhiwei Dong, Yifan Pu, Yongqiang Yao, Bo Li, Ruihao Gong, Haoran Wang, Xianglong Liu, Gao Huang, and Wei Wu. The first four made equal contributions, and Ruihao Gong was the project leader.

  2. AidgetRock - Score: 36.972, Accuracy: 0.55421, Time: 15 ms.
    The team members are Zifu Wan, Xinwang Chen, Ning Liu, Ziyi Zhang, Dongping Liu, Ruijie Shan, Zhengping Che, Fachao Zhang, Xiaofeng Mou, and Jian Tang. The project leader was Ning Liu.

  3. ENOT - Score: 8.974, Accuracy: 0.60089, Time: 67 ms.
    The team members are Alexander Goncharenko, Max Chuprov, Andrey Sherbin, Sergey Alyamkin, and Ivan Malofeev.

Competition Progression

Below is displayed the progression of the submissions over the course of the competition. As mentioned above, there was a power outage on 07/10, causing the lack of data points around those dates.

Score ProgressionAccuracy ProgressionTime Progression


Special Interest Group on Design Automation


NameRole and OrganizationEmail

Benjamin Chase Boardley

Student, Purdue University


Ping Hu

Student, Boston University


Yung-Hsiang Lu

Professor, Purdue University


Gowri Ramshankar

Student, Purdue University


Kate Saenko

Professor, Boston University


Nicholas Synovic

Student, Loyola University Chicago


George K. Thiruvathukal

Professor, Loyola University Chicago


Oscar Yanek

Student, Loyola University Chicago


Leo Chen

Student, Purdue University