Attachment | Size |
---|---|
![]() | 234.42 KB |
We are delighted to launch the 1st Scientific Figure Captioning (SciCap) Challenge!
This competition calls researchers specializing in AI, NLP, and CV to design systems capable of generating captions for all types of scientific figures in arXiv papers. This challenge will be hosted at the CLVL workshop in ICCV 2023.
We encourage everyone interested to take part in this exciting venture. To join us, please visit: http://SciCap.AI
== Challenge Overview ==
The challenge consists of two stages: the Test Phase and the Challenge Phase.
- During the Test Phase, which lasts approximately 2.5 months, each team uses the provided training set, validation set, and public test set to build and test their models.
- Then, in the Challenge Phase, which takes place in the final 2 weeks before the submission deadline, we will release a hidden test set of a similar size to the public test set. Teams can then submit their results for this hidden test set.
The winning teams will be determined based on their best results from the hidden test set.
== Important Dates ==
- SciCap Challenge launches: May 29, 2023
- Test Phase begins: May 29, 2023, 12:00 AM (CST)
- Challenge Phase begins (hidden test set released): August 15, 2023, 12:00 AM (CST)
- Challenge Phase ends: August 31, 2023, 23:59 PM (CST)
- Technical report submission deadline: September 4, 2023
- Challenge result notification: September 6, 2023