Sign Language Translation (SLR) is the task of automatically recognising signs from video sequences of sign language. It has significant importance as it can provide a means of communication for people with hearing or speech impairments, as well as facilitating communication between hearing and deaf communities.
Auslan is the sign language used by Australian Deaf and Hard of Hearing (DHoH) community. Coding Fest 2025 organises an AI for Auslan challenge to seek innovative AI solutions to help better communication between DHoH community and the hearing community.
This is a great opportunity to strengthen and practice your AI skills for social good. There may also be an opportunity for academic publications.
(We use the performance on private test set for ranking.)
The data we use is from a TV series named ‘Sally & Possum’. Australian Sign Language (Auslan) is used in this show. (
https://iview.abc.net.au/show/sally-and-possum)
There are 13214 videos included in the training dataset. You
need to split the training and validation set if necessary. The subtitle for each clip and ‘video-clip-name’ are provided inside the .csv file.
video-clip-name: str, Clipped video ID.
subtitle: str, Auslan of clipped video.
The dataset can be downloaded in Google Drive: https://drive.google.com/drive/folders/1w_MOEjqTbI6DZB03c_vxoaGXNaF4GB06
For any enquiries, please contact: codingfest.top@gmail.com or AI Auslan Organization Team members.
In order to ensure fair competition, each participant is limited to a maximum of five submissions per day. The leaderboard will be updated on the following day.
# | Team Name | BLEU@4 Score |
---|---|---|
1 | Don't Overfit | 5.375 |
2 | LittleHelper | 5.330 |
3 | NeuroSigns | 4.906 |
4 | Little rednote | 4.896 |
This competition challenges participants to develop models effective in Australian Sign Language (Auslan) translation in a novel dataset, while leveraging the valuable training data provided by the Auslan-Daily dataset.
For the evaluation of the task of Sign Language Translation, we will report the BLEU@4 score on the testing set of the Auslan-Daily dataset.
@misc{ title={Auslan-Daily: Australian Sign Language Translation for Daily Communication and News}, author={Xin Shen, Shaozu Yuan, Hongwei Sheng, Heming Du, Xin Yu}, year={2023} }