新闻
Studies on Sign Language Retrieval in Low-Resource Dictionary Settings
arXiv cs.AI · 发布于 · 阅读约3分钟
30秒读懂
- 发生了什么
- Researchers trained a pose encoder on American Sign Language and tested it on six other sign languages without retraining, achieving 36-81% retrieval accuracy with minimal video examples.
- 为何重要
- Matters for engineers building sign language tools for under-resourced languages where annotated data and large corpora do not exist.
- 注意
- Performance drops significantly on larger dictionaries; the approach works best with one to four video examples per sign, limiting real-world applicability.
- retrieval
- encoder
- eval
这条新闻背后的模式
每个模式都讲清楚技术如何运作、何时值得投入，以及在哪里会失效。
The Agent Architect
每周一个模式、一个权衡、一个生产事故案例。为构建智能体系统的人准备的每周简报。
每周一封邮件,一键退订。您的地址仅用于发送简报。