Hello TrojAI participants,
The object-detection-feb2023 round is now open on the leaderboard:
https://pages.nist.gov/trojai/Data download is currently on a per-request basis; just while we move the dataset through the publication and release process.
https://twitter.com/iarpanews/status/1634196571741999104object-detection-feb2023 looks at trojaning object detection AI system where there is higher model spare capacity compared to models trained on the COCO dataset. Refer to the data page for more details.
Make sure all submissions base their container on the trojai-example found at
https://github.com/usnistgov/trojai-example/tree/object-detection-feb2023If you have not submitted a container to the TrojAI evaluation system in awhile, please consider doing so.
We look forward to seeing everyone's performance.
Good luck trojan hunting.
Regards,
The TrojAI Team