Dear all,
We are pleased to invite you to participate in the 2025 SignforAll Challenge, which will be featured in the IEEE International Workshop on Machine Learning for Signal Processing (MLSP) 2025. This shared task is part of the special session "Sign Language Translation in the Era of Large Language Models - Beyond English", focusing on advancements in sign language translation, particularly the integration of large language models and moving beyond English-centric approaches. We also welcome papers about sign language translation outside the challenge.
This special session addresses critical challenges at the intersection of AI and accessibility. It introduces a novel dataset co-developed with sign language users and interpreters, providing a platform to explore the generalization capabilities of sign language translation models—one of the fundamental limitations in the field.
Participants will develop models to translate SSL signs into text across three challenging tracks:
Track 1 - Model Generalization on Unseen Signers: Recognize SSL signs performed by unseen signers but using seen sentences.
Track 2 - Model Generalization on Unseen Sentences: Recognize SSL signs performed by seen signers but with previously unseen sentences.
Track 3 - Model Generalization on Unseen Signers and Sentences: Recognize SSL signs performed by fully unseen signers and with unseen sentences.
The dataset and baseline code will be available on Kaggle and GitHub.
For inquiries, please contact us at: signf...@googlegroups.com.
We look forward to your participation and contributions to advancing sign language translation research!
Best regards,
Ahmed
On behalf of Sign Language Translation Challenge & Special Session (MLSP 2025)