day
hr
min
sec
participants / team
Advising Organization:Ocean Affairs Council
As marine debris becomes a critical global issue, leveraging AI technology to enhance environmental monitoring is a major international trend. To accelerate the digital transformation of marine governance, this "International AI Challenge for Marine Debris Image Recognition" will release a high-quality marine debris image dataset. We invite outstanding professionals from global academia, research institutions, industry, and tech communities to develop high-precision object detection models. Your participation will drive innovation in marine monitoring and promote practical AI applications in environmental governance.
This competition is a vital platform connecting global innovation. Whether you have a background in AI, data analysis, computer vision, marine science, or are simply passionate about sustainability, you are welcome to join. Teams will not only have the opportunity to win prize money and gain international exposure but will also use real-world data to transform technical innovation into tangible environmental impact. Join us in tackling marine sustainability challenges with AI and building a smarter future together!
Teams ranked in the top 10 on the Private Leaderboard with an mAP@0.5 score of 0.6 or above will qualify for the Final Round. The Final Round offers a total prize pool of up to NT$300,000.
All event times are based on Taiwan Time (GMT+8).
| Date | Event |
|---|---|
| 2026/07/09 | Registration Opens |
| 2026/07/17 12:30~13:30 | Online Information Session |
| 2026/07/29 12:30~13:30 | Online Information Session |
| 2026/08/10 23:59:59 | Registration Deadline |
| 2026/08/15 | Preliminary Technical Workshop |
| 2026/08/15 09:00:00 | Test Set Released & Available for Download |
| 2026/08/15 09:00:00 | Submission for Preliminary Answers |
| 2026/09/01 23:59:59 | Deadline for Preliminary Submissions |
| 2026/09/03 | Announcement of Finalists |
The scoring rules for each query game set (one question) are as follows:
Exact playing-strength prediction (Top-1): 1 point
Playing-strength error of ±1 level (but not exactly correct): awarded ${1}/{e} \approx 0.368$ points
All other cases: 0 points
$$score_{i}=\begin{cases} 1, \text{exact rank match}\\ \frac{1}{e}, \text{within} \pm1 \text{ rank (excluding exact)}\\ 0, \text{otherwise} \end{cases} $$
$$S_{strength}=\frac{1}{N}\sum\limits_{i=1}^{N}{score_{i}}$$