2020 Low-Power Computer Vision Challenge

Announcement of Winners

More Info

FPGA Track

This track requires participatants to recognize objects in images using a field programming gate array. Nineteen teams submitted 115 solutions.

PyTorch Video Track

This track requires that participating teams recognize English letters or numbers in the signs in the video captured by an unmanned aerial vehicle (UAV). Eleven teams submitted 84 solutions to this track.

1OnceForAllMIT HAN Lab
2MAXXNational Chiao Tung University, Taiwan
3WaterSKLCA, Institute of Computing Technology, CAS
1LPNetByteDance Inc
2TAMU-KWAITexas A & M University and Kwai Inc
3C408Stony Brook University (SUNY Korea)

OVIC Object Track

This track requires that participating teams perform interactive object detection on COCO detection models operating at 30 ms / image on a Pixel 4 smartphone (CPU).

OVIC Image Track

This track requires participants to perform Real-time image classification on Imagenet classification models operating at 10 ms / image on a Pixel 4. Sixteen teams submitted 179 solutions to both these OVIC tracks.

1OnceForAllMIT HAN Lab
2SjtuBicaslShanghai Jiao Tong University
3NovautoNovauto Inc
1BAIDU&THUBaidu Inc & Tsinghua University
2imchinfeiBeihang University
3LPNetByteDance Inc


This track focuses on Imagenet classification models operating at 7 ms / image on a LG G8 smartphone (DSP). Participants are required to provide the model with the highest top-1 accuracy to win awards.

1LPNetByteDance Inc
2imchinfeiBeihang University
3 (Tie)FoxPanda; OnceForAllNational Chiao Tung University and National Tsing Hua University; MIT HAN Lab


The Low-Power Computer Vision Challenge is an annual competition started in 2015.


Computer vision technologies have made impressive progress in recent years, but often at the expense of increasingly complex models needing more and more computational and storage resources.

Our Aim

This workshop aims to improve the energy efficiency of computer vision for running on systems with stringent resource constraints.

Our Plan

This workshop will discuss the state of the art of low-power computer vision, challenges in creating efficient vision solutions, promising technologies that can achieve the goals, methods to acquire and label data, benchmarks and metrics to evaluate progress and success.


  • All
  • 2015
  • 2016
  • 2017
  • 2018
  • 2019


*Any opinions, findings, and conclusion or recommendations expressed in this material are those of the organizers and do not necessarily reflect the view of the sponsors.

Organizing Committee

Started by Professor Yung-Hsiang Lu from Purdue University in 2015.

Mohamed M. Sabry Aly

IEEE Council on Design AutomationNanyang Technolgical University

Ming-Ching Chang

University at Albany - SUNY

Bo Chen


Shu-Ching Chen

Florida International University

Yiran Chen

ACM Speical Interest Group on Design AutomationDuke University

Xiao Hu

manage lpcv.ai websitePurdue student

Jaeyoun Kim


Mark Liao

Academia Sinica, Taiwan

Tsung-Han Lin


Yung-Hsiang Lu


An Luo


Naveen Purushotham

Xilinx Inc.

Tim Rogers


Mei-Ling Shyu

IEEE Multimedia Technical CommitteeMiami University

Joseph Spisak


George K. Thiruvathukal

Loyola University Chiicago

Anirudh Vegesana

manage lpcv.ai websitePurdue student

Carole-Jean Wu


Junsong Yuan

University at Buffalo

Contact Us


Yung-Hsiang Lu
School of Electrical and Computer Engineering
Electrical Engineering Building
465 Northwestern Ave
West Lafayette, IN 47907-2035

Phone Number

+1 765 494-2668