OpenCV AI Competition 2026, powered by AWS

Overview

The OpenCV AI Competition 2026, powered by AWS, invites participants to develop innovative vision systems leveraging OpenCV 5 and Amazon Web Services. The competition emphasizes Physical AI and Generative AI systems where visual understanding drives practical decisions, actions, predictions, or interactions. Submissions are expected to demonstrate measurable value and be shared with the global OpenCV community. Participants are required to use OpenCV 5 for substantive image or video analysis, with a meaningful component running on AWS. The competition offers two optional featured paths: the Best Use of COOL Award and the Agentic Vision Award, each with a separate prize. Teams can pursue either path, combine both, or take another approach while remaining eligible for general prizes. The competition timeline includes a registration and proposal submission phase running from August 12-2026, with proposals accepted starting August 13. Fifty teams will receive a cloud compute grant valued at $150. All other teams are still eligible to proceed to the two-month build phase, which takes place from August 26 through October 26, 2026. Final projects are due on October 26, 2026, at 11:59 p.m. Pacific Time. Judging will occur from October 27 through November 9, 2026, with winners announced on November 10, 2026, during an OpenCV Live! webinar or another announced OpenCV channel. The total prize pool is $20,250 in cash, distributed across grand prizes and special category awards, alongside compute grants for selected teams.

What to build

Participants are to build vision systems that see, reason, and act using OpenCV 5 for substantive image or video analysis, with a meaningful component running on Amazon Web Services (AWS). The competition encourages projects in Physical AI and Generative AI systems where visual understanding drives useful decisions, actions, predictions, or interactions. Projects can pursue either the Cloud-Optimized Vision with COOL path, the Agentic Vision path, both, or another approach.

Prizes

Key dates

• Competition launch, registration, and proposal submissionAugust 12, 2026
• Proposals accepted startingAugust 13, 2026
• Build PhaseAugust 26 - October 26, 2026
• Zoom check-in for compute grant recipientsSeptember 21 - October 2, 2026
• Final projects dueOctober 26, 2026, 11:59 p.m. Pacific Time
• Final JudgingOctober 27 - November 9, 2026
• Winners announcedNovember 10, 2026

How to enter

1. Register for the competition.
2. Submit a proposal (optional, for AWS compute grant consideration).
3. Build, evaluate, document, and submit a final project between August 26 and October 26, 2026.
4. For compute grant recipients, complete a 30-minute Zoom check-in between September 21 and October 2, 2026.

Submission requirements

• Every entry must use OpenCV 5 for substantive image or video analysis.
• A meaningful component of the project must run on AWS.
• For AWS Grant Proposals: Include a team name, clear problem statement and intended real-world impact, planned OpenCV 5 image or video analysis, planned AWS architecture and services, high-level architecture diagram or technical description, target users or beneficiaries, proposed evaluation method and judge demonstration, intention to pursue COOL/Agentic Vision path(s) or neither, and a short team bio.
• Final Submission Requirements: A technical report, a public or private judge-accessible code repository or archive, pinned dependencies plus clear build/deployment/test instructions, an architecture diagram showing OpenCV 5 and AWS components, a working web endpoint or arranged live screen-share demonstration, a public or unlisted judge-accessible video (max 5 minutes), and evaluation evidence appropriate to the project.
• For Best Use of COOL Award: COOL must execute the claimed core workload on AWS Graviton or the Arm component of a documented hybrid architecture. Documentation of reproducible deployment and comparison against an appropriate baseline is expected. Additional evidence includes COOL version, AWS instance/deployment configuration, reproducible evaluation method/inputs/baselines/results, and evidence that COOL executes the claimed core workload.
• For Agentic Vision Award: Build a workflow where an agent uses OpenCV 5 tools within a multi-step perception-decision-action loop, where image or video results influence a subsequent plan, tool call, action, or request for human approval. Additional evidence includes an agent workflow diagram, a trace or demonstration showing OpenCV 5 output changes a later decision/tool call/action, and evaluation of task success/failure handling/observability/human control.

Judging criteria

• Technical execution (30%): correctness and depth of the OpenCV 5.
• Innovation (20%): originality and thoughtful use of computer vision or AI.
• Real-world impact (20%): importance of the problem, usefulness, and evidence of potential benefit.
• User experience (10%): usability, accessibility, and quality of interaction.
• Documentation and presentation (10%): clarity of the report, code, architecture, instructions, and video.
• Cloud delivery, reproducibility, and responsible operation (10%): AWS deployment quality, repeatability, observability, security, and responsible-use practices.

Sponsors & partners

Resources

• OpenCV AI Competition 2026 Devpost Page