Introduction
Lip reading aims to recognize spoken content based solely on visual information derived from the speaker’s lip movements. This emerging and challenging field lies at the intersection of computer vision and natural language processing and plays a key role in various applications in different domains. In the previous Chat-scenario Chinese Lipreading (Chat- CLR) challenge at ICME 2024, we focused on realworld home conversations and set up two tasks: Wake Word Lipreading and Target Speaker Lipreading. The challenge attracted 23 teams to download the dataset, with 11 teams submitting their results. Ultimately, 6 papers were accepted into the ICME 2024 proceedings and presented orally during the Grand Challenge session.
Meetings represent one of the most valuable yet challenging contexts for lipreading due to the rich information exchange and decision-making processes involved. The meeting-scenario Chinese Lipreading (MeetCLR) challenge centers on the multi-speaker lipreading task, where both the training and evaluation datasets involve different groups of speakers.The specific tasks are as follows:
1) Lip-reading Speaker Diarization.
2) Lip-reading Speech Recognition.
The following resources will be provided:
- A large-scale Chinese lipreading corpus and a comprehensive baseline system.
- A public benchmark for fair comparisons.
- A challenge session to foster communication.
- A summary publication highlighting the most effective techniques and promising directions.
Planned Schedule (AOE Time)
- Registration opens: December 21, 2024
- Training set releases: December 27, 2024
- Development set and baseline system release: January 7, 2025
- Registration closes, evaluation set release and leaderboard update for evaluation: February 10, 2025
- Leaderboard freeze: March 15, 2025
- System report submission: March 22, 2025
- Final paper submission: March 25, 2025
Organizers
Contact Us
For additional information, please email us at MeetCLRchallenge@gmail.com.