LOOKOUT challenge: MIT-licensed starter preview

The first public preview of the LOOKOUT maritime monocular visual odometry challenge is available.

Start with the repository README to run the synthetic evaluator example on CPU, inspect the prediction format and prepare the pinned DPVO inference environment. This preview includes the starter source, local evaluator, development rules and setup guide.

Real RGB/thermal recordings, reviewed timing/calibration/reference downloads and the reproduced public-data baseline will follow. The synthetic example is artificial and is not a real-data result. Competition dates, online submissions and the leaderboard will be announced separately. The starter code is MIT-licensed. Dataset and upstream dependency/checkpoint terms are separate.

• Challenge page: https://macvi.org/workshop/macvi27/challenges/lookout
• Start here: https://github.com/LOOKOUT-AI/lookout-vo-challenge#try-the-evaluator-now
• Progress: https://github.com/LOOKOUT-AI/lookout-vo-challenge/blob/main/RELEASE_STATUS.md

Watch releases for dataset updates. Feedback on the starter and interface is welcome in GitHub issues.

This licensed preview supersedes v0.1.0-preview.1.

LOOKOUT challenge: public starter preview

This initial source preview has been superseded by the MIT-licensed starter preview:
https://github.com/LOOKOUT-AI/lookout-vo-challenge/releases/tag/v0.1.0-preview.2

Use that release or the current main branch for the starter, synthetic CPU evaluator example and setup guide. Real dataset downloads and the public-data baseline remain forthcoming. Dataset and upstream dependency terms are separate from the starter code license.