AICLD 人工智能辅助增量式中文唇语识别数据库平台 AICLD Platform for AI-Assisted Incremental Chinese Lip-Reading Database
- Led architecture and full-stack development of the AICLD lip-reading database platform, building a Streamlit web portal unifying data indexing, versioning, and task management. Designed and deployed a cloud-backed fully automated incremental acquisition pipeline with annotation/QA workflows and real-time pipeline monitoring for daily corpus updates.
- Authored full technical documentation and user manuals defining the automated acquisition pipeline, data-request protocols, and support channels. Curated the literature library and user-feedback mechanisms to enable efficient, compliant access to the large-scale corpus for research teams.
AICLD is a Mandarin lip-reading database platform accessible in mainstream browsers, with cloud-backed fully automated incremental corpus growth. The sample count is already large and grows daily, poised to surpass comparable public corpora. The portal combines dataset overview, detail lookup, documentation of the automated acquisition pipeline, literature search, and user feedback. Data are not open-sourced yet and can be requested by email; an online lip-reading service is planned, together with usage policies and technical support channels.
