Low-Power Image Recognition Challenge (LPIRC 2019)

 

First Announcement and Call for Participants:
2019 Low-Power Imaging Recognition Challenge (LPIRC 2019)

Held as part of Computer Vision and Pattern Recognition Conference (CVPR 2019)

Long Beach, CA, USA

 

 

 

Previous LPIRC Editions:

 

CVPR Workshop on Low-Power Image Recognition Challenge

Date: 2019/06/16 (or 17, to be finalized)
Location: Long Beach, California
Further information on registration will be available soon.

This workshop will extend the success of LPIRC in the past four years identifying the best vision solutions that can simultaneously achieve high accuracy in computer vision and energy efficiency. Since the first competition held in 2015, the winners’ solutions have improved 24 times in the ratio of accuracy divided by energy.

LPIRC 2019 will continue the online + onsite options for participants. One online track with pre-selected hardware allows participants to submit their solutions multiple times without the need of traveling. One onsite track allows participants to bring their systems to CVPR. Two tracks in 2018 will be offered using the same setup and rules. Both tracks are object detection.

Track 1 (online): Tensorflow Model. Submission dates: 2019/05/15-2019/06/10
Track 2 (onsite): No restriction. Bring your system (no hardware or software restriction) to CVPR.

For Track 2, Xilinx Ultra96 + PYNQ framework http://www.pynq.io/community.html is an option available to participants. Xilinx will provide support and possible HW donation. People interested in this opportunity may contact Naveen Purushotham (npurusho@xilinx.com) for more details.

All participants must register before 2019/05/15.

To win a prize in 2019 LPIRC, a solution must be better than the state of the art. Please notice that this is a new requirement in 2019. The requirement in each track is shown below. If a solution’s score is below the number, it cannot win. If a solution’s score is better, it still has to beat the other solutions submitted to 2019 LPIRC.

Track 1 (online): mAP 0.14684
Track 2 (onsite): mAP / energy (WH) 0.4446

LPIRC 2019 is part of the Embedded Vision Workshop in CVPR. LPIRC 2019 will feature three speakers, a session on future low-power computer vision, and the onsite competition. The current program for the LPIRC Workshop is shown below.

5 minutes Welcome by Organizers
40 minutes Speech by the Competition Winners (2018 LPIRC-II)
  • Amazon: Tao Sheng
  • Expasoft: Alexander Kondratyev
60 minutes Invited Speakers: Kurt Keutzer (Berkeley)
50 minutes Grand Challenge on Low-Power Computer Vision (call for paper below)
60 minutes On-site competition

 

Call for Papers: Grand Challenge in Low-Power Computer Vision

Competitions are widely adopted in academia, industry, and government to propel technologies forward. Well-known competitions include the DARPA Autonomous Vehicle Challenge and the Space X prize. The competitions have been attributed as the accelerators of making significant progress in the technologies. The CVPR Workshop on LPIRC solicits papers that describe future competitions of low-power computer vision. Authors are encouraged to think boldly, imagining “Grand Challenge” or “X Prize” type of competitions. The paper should describe one (or several) competitions that are (1) easy to understand, (2) easy to evaluate success objectively, (3) possible but at least 5 years beyond today's technologies.

Authors are encouraged to discuss the following issues in their papers:

  • Description of the data (e.g., image, video, with or without caption)
  • Is any existing dataset used? If so, is the dataset sufficient or additional improvements are needed?
  • Methods to acquire the data
  • Description about copyright and privacy (if applicable)
  • Methods to annotate the data (if applicable, for getting the ground truth)
  • Methods for scoring
  • Infrastructure for scoring (if applicable, for example online grading server or harness)
  • Constraints of hardware (if applicable, for example weight, size)
  • Minimum or maximum requirements (e.g., accuracy, energy. execution time)
  • Differences from existing competitions

All submissions must be in PDF format. Submissions are limited to ten pages, including all figures, tables, and references, in the CVPR 2019 submission style. Further details on submission will be available soon.

Selected proposals will be presented at the LPIRC workshop at CVPR 2019.

 

Sponsors

IEEE Rebooting ComputingGoogleIEEE Circuits and Systems SocietyXilinxMediaTek

 

LPIRC 2019 is sponsored by Xilinx, Google, Mediatek, IEEE Circuits and Systems Society.

LPIRC 2015-2018 sponsors include Google, Facebook, Nvidia, Xilinx, Mediatek, IEEE Circuits and Systems Society, IEEE Council on Electronic Design Automation, IEEE GreenICT, and IEEE Council on Superconductivity. IEEE Rebooting Computing is the founding sponsor of LPIRC.

Sponsorship Opportunities Available