Challenge schedule

04/21/2022 Start of the challenge (development phase)

Release of training (with ground truth) and validation data (without ground truth).

Time zone: Codalab will use UTC for all competition tracks/phases.

06/15/2022 Release of test data (without ground truth) and validation labels

Release of test data (without ground truth) and validation labels.

At this stage, parcitipants can download the test data and validation labels, both encrypted. The encryption key will be released in a few days ahead. With the encryption-decryption strategy, we ensure participants with lower speed Internet connection are at no disadvantage.

06/17/2022 Start of test phase (end of development phase):

Release of test data (and validation labels) decryption keys. Participants start predicting the results on the test data. At this stage, they can retrain (or fine-tune) their models with the additional validation data/labels.

06/24/2022 End of the Challenge

Deadline for submitting the final predictions over the test (evaluation) data.

06/26/2022 Code and fact sheets submission

  • Deadline for code submission with detailed instructions (known as “code verification stage”). This includes a list of requirements, pre-trained models, and so on. Note, training code with instructions is also required. Organizers strongly encourage the use of docker to facilitate reproducibility.
  • In addition to the code, participants are requested to submit the fact sheets using a template provided by the organizers.

Deadline timezone: Anywhere on Earth Time Zone - AOE

Additional details about code and fact sheets submission are provided in our challenge webpage, Wining solutions (post-challenge) section.

07/01/2022 Release of final results

We encourage participants to submit a paper to the associated workshop, independently of their rank position.


ECCV2022 Challenge on Sign Spotting

The ChaLearn ECCV2022 Challenge on Sign Spotting is open and accepting submissions on Codalab. Train data is available for download. Join us to push the boundaries of Continuous Sign Language Recognition.