【复旦大学研发“眸思”大模型，为视障者开启“看见”新途径】
据复旦大学官方公众号最新报道，该校自然语言处理实验室（FudanNLP）的科研团队成功研发出名为“眸思”（MouSi）的多模态大模型，并以此为基础打造了一款名为“听见世界”的应用程序，旨在帮助视障人士更好地理解并感知周围环境。
这款创新的“听见世界”App，凭借一枚普通的摄像头和一对耳机，就能将实时画面转化为语音描述，为视障者构建起一个声音版的视觉世界。该系统不仅能够精准地将图像信息转化为语言，如识别人物、物体和文字，还具备描绘场景、预警潜在风险等先进功能。例如，当视障者面临障碍物或者交通危险时，App将及时发出语音提示，保障他们的行动安全。
“眸思”大模型的诞生，是人工智能技术在无障碍领域的一次重要突破，它将科技的力量注入到公益之中，为视障人群的生活带来革命性的改变。复旦大学FudanNLP团队的这一创新成果，无疑为全球无障碍技术的发展树立了新的标杆，也为构建包容性社会贡献了科技智慧。
这一创新项目已上线并投入使用，期待它能为更多的视障人士打开一扇新的“视”界之窗，让他们能够更好地“看见”并融入这个世界。
英语如下：
**News Title:** “Fudan University Develops ‘MouSi’ Model, Empowering Visually Impaired to ‘Hear’ the World via App”
**Keywords:** MouSi Model, Visual Impairment Assistance, Hear the World
**News Content:**
**Fudan University’s “MouSi” Large Model Paves a New Path for the Visually Impaired to “See”**
According to the latest report on Fudan University’s official WeChat account, the research team from the university’s Natural Language Processing Laboratory (FudanNLP) has successfully developed the “MouSi” multimodal large model, which forms the basis for an application called “Hear the World,” designed to aid the visually impaired in better understanding and perceiving their surroundings.
This innovative “Hear the World” app, utilizing a standard camera and a pair of headphones, can convert real-time images into voice descriptions, creating an auditory version of the visual world for the visually impaired. The system not only accurately translates image information into language, recognizing people, objects, and text, but also boasts advanced functions like scene depiction and potential risk预警. For instance, when a user approaches an obstacle or a hazardous traffic situation, the app promptly provides audio alerts to ensure their safety.
The birth of the “MouSi” model marks a significant breakthrough in the application of artificial intelligence (AI) technology in the field of accessibility. It harnesses the power of technology for the greater good, bringing a revolutionary change to the lives of visually impaired individuals. FudanNLP team’s innovative achievement sets a new benchmark for global accessibility technology and contributes technological wisdom to building an inclusive society.
This pioneering project has already been launched and is in use, with expectations that it will open a new “window” to the world for many visually impaired individuals, enabling them to better “see” and engage with the world around them.
【来源】https://www.ithome.com/0/753/295.htm
Views: 1