fbpx

OpenCV Spatial AI Contest

OpenCV is excited to partner with Intel and Microsoft Azure to bring an AI, computer vision, and manufacturing competition to life!

* Due to global shipping issues and the hard deadlines necessary for the Contest we had to make the hard choice to limit the scope of the Contest to these regions.
Sponsored by:
The Intel logo, and other Intel marks are trademarks of Intel Corporation or its subsidiaries.

Grand Prize Winners

Congratulations to the very deserving Grand Prize Winners, as chosen by our judges

#1 EyeCan Reborn

ScanNERF - Neural Radiance Field Capture

#2 UCA Robotics

Vision Based Human-Robot Interaction

#3 Potato Lovers

Colorado Beetle Buster

Winners Announcement

From OpenCV Weekly Webinar

Popular Vote Winner

Chosen by fan vote!

B-Arol-O

ARNEIS

Introduction

Welcome to the OpenCV Spatial AI Contest, sponsored by Intel and Microsoft Azure. The previous two OpenCV AI competitions (2020, 2021) have been amazing, but this one has a special twist. This contest tasks teams with creating a replica of a real-world manufacturing solution using LEGO blocks and OpenCV AI Kit – Lite (OAK-D-LITE), a tiny yet powerful edge AI camera engineered by Luxonis and capable of neural network processing plus so much more.

Thanks to the generosity of our sponsors we are able to provide 50 teams with the resources to help create something amazing over the 3-month build phase.

A good submission will include your problem statement, proposed solution, and the LEGO kit(s) needed to build it! The submission phase is open until November 22nd, 2021 with the Phase 1 winners being announced in December.

From there teams will build starting January 1st until the final submission deadline in April. See the full schedule.

To our wonderful community- thank you, and we can’t wait to see your unique project submissions!

List of Finalist Teams

The following teams and projects were selected from over 120 submissions by a panel of expert judges. Congratulations to these teams, a deep thank you to all who submitted, and best of luck in the build phase!

Team NameRegion
2+2=5(Pseudo Boom)Europe
B-AROL-O Bottling SystemsEurope
Brown BeltNorth America
Buckeye WildcatsNorth America
Eclectic EdmontoniansNorth America
Education EmpowersNorth America
EPFL LoutresEurope
Evtek electrified ecological envoysNorth America
Eyecan RebornEurope
Feel the BurnJapan
Fire Rabbits TeamNorth America
French Fries FiestaEurope
Getafe Poka-yokersEurope
Greenmatics
Australia & New Zealand
Guadalquivir TitansEurope
Hell CleanerEurope
I-AM realEurope
ICU Vision AIEurope
Kiev Robo ArtEurope
Kiwi Bot Builder
Australia & New Zealand
Kleve-GoGettersEurope
LEGO BearsNorth America
Lone Peak LevelersNorth America
Los Angeles InoculumNorth America
LoughBrosEurope
MAGRITONorth America
Maryland TerpNorth America
Mexican FlowNorth America
Midwest DefendersNorth America
Moscow RobophobesEurope
Nature AvengersNorth America
Northern Ireland DigitalizationEurope
Ocean Crusaders
Australia & New Zealand
PablitoEurope
Potato loversEurope
Prairies GrasslandsNorth America
QUALITYFIRWALLEurope
Regina Lego MavericksNorth America
Silesian CarbonsEurope
SoFlo Primary KeyNorth America
Team AniBOT
Australia & New Zealand
Team DeodatesEurope
Team WahWahTronEurope
The SourdoughersNorth America
Team TorontoNorth America
UBCO AR SolutionsNorth America
UCA RoboticsEurope
US-CobotsNorth America
Wavicle DynamicsNorth America
Wro Botic ElephantEurope

The OAK-D-LITE Device

200 OAK-D-LITE devices will be given to phase one winners to help them complete their projects. The OAK-D-LITE is a variant of the OpenCV AI Kit (OAK) capable of Spatial AI.

What is Spatial AI? It’s the capability for AI to be applied to the physical world – to tell you what an object is and where it is in 3D space – in real time.

OAK does this by running object detection with its integrated 12MP RGB camera and combining the results with its integrated stereo-depth engine. You can run a variety of deep learning models like those supported by OpenVINO and OAK-D-LITE automatically augments them with spatial data from the integrated stereo depth engine.

Prizes

$45,000 in prizes, sponsored by Intel and Microsoft Azure

A Creator Box including the following will be given to the top 50 teams.

Additional Prizes will be awarded to top completed projects.

Global Prizes will be awarded to the top projects across all regions:
Winning teams will be featured at Microsoft Build with a professionally-produced video courtesy of Intel.

Contest Timeline

Phase 1 : Selection

The competition will consist of two phases; selection and finals. Starting November 2nd 2021, teams may apply for the competition with a project proposal. The submission period will end on November 22nd 2021 at 11:59pm PST. Proposals must include the following important sections:

  •  Team Name : We encourage people to use creative team names that include their University, Company, or City name + some creative phrase. E.g. San Diego Dolphins, or the Cambridge Turings.
  • Team Capability : A full list of team members and qualifications of the team.
  • Problem Statement : A description of a real-world replica of a manufacturing problem and proposed solution using OAK-D-LITE and the necessary LEGO kits (up to $400) within 3 months of work. The OAKs and LEGOs will be provided by our sponsors.
  • Region: Where the team is located.

Phase 2 : finals

The 50 teams selected for the finals have 3 months to complete their project.  Finalist teams will be announced December 2021.  Finalists will be awarded a Creator Box to accomplish their projects, which will be mailed December 2021. Finalists will have from January 1st until April 4th 2022 at 11:59pm PST to complete their projects and submit final reports. Grand Prize Winners will be announced April 21st, 2022.

One (1) representative of each winning team will receive an expense paid trip (an economy flight ticket, per-diem accommodation, and hotel for up to $4k) to present their solution at Microsoft Build! (Pandemic restrictions permitting)

Submission Guidelines : Do's and Don’t

The organizers of the competition believe AI is like electricity with the potential to disrupt and transform many areas of our lives. It will have a positive impact on a wide variety of industries like autonomous driving, healthcare, agriculture, manufacturing, assistive technologies, entertainment, safety and security to name a few. We encourage submissions in all these areas. However, AI can also be used in ways that are unethical and/or discriminatory.

Participants should review the TensorFlow Responsible AI page as well as the Microsoft Responsible AI Resources page before submitting a proposal. See the Sample Submission PDF for how to format your submission.

Frequently Asked Questions

You've got questions, we've got answers.

  • Integrated 4K 13 MP Color Camera
  • JPEG, H.264 & H.265/HEVC Encoding
  • Integrated 480 P Global Shutter Synchronized Stereo Pair
  • USB 3.0 Type-C Interface
  • Use with any host OS that OpenVINO supports

  • Stereo Min Distance: ~18cm (using extended disparity search stereo mode)
  • Stereo Max Distance: ~18m
  • Stereo Camera Hyperfocal Distance: 6.5cm
  • Stereo Camera Baseline: 7.5cm
  • Stereo Horizontal Field of View (HFOV): 72.9 degrees
  • Stereo Camera Resolution: 640×480
  • RGB Camera resolution: 4208×3120
  • RGB (AutoFocus) Min Distance: 8cm

Teams are allowed 1-3 members.

Choose the region that most team members are located in, and that your solution focuses on. Use your best judgement.

Yes. See the documentation for examples showing how to run OpenVINO models.

You can also run your own (custom-trained) models, learn how in the tutorial.

Yes! Send an email to [email protected] with your submission formatted as a PDF.

OAK-D-LITE’s provided software is officially available for Python and C++, with a community-support Java version.

We prefer a PDF. You can see an example of what we will expect here: Download Sample Application PDF

 

The deadline for submissions is November 15th, 2021 at 11:59pm PST.

Yes! You can resubmit your project any time up to the deadline which is November 22nd, 2021 at 11:59pm PST. The original deadline of November 15 was extended.

Yes, teams can be just one person, but it’s much more fun to collaborate!

We will contact the winners via email in December 2021 which will be followed by an announcement on this page.

Yes! We love videos. Please send any video links inside your proposal.

Your project can use any license you choose. Closed-source projects will be required to provide extensive live demonstrations to judges.

Unfortunately due to global shipping issues and needing to have hard deadlines for the various Contest milestones we had to make the hard choice to limit the scope of the Contest to those regions.

Team leads have been emailed instructions as of March 20th, 2022. If you have not received this email, please contact us at [email protected]

Subscribe and Start Your Free Crash Course

Stay up to date on OpenCV and Computer Vision news and our new course offerings

Join the waitlist to receive a 20% discount

Courses are (a little) oversubscribed and we apologize for your enrollment delay. As an apology, you will receive a 20% discount on all waitlist course purchases. Current wait time will be sent to you in the confirmation email. Thank you!