星际大模型落地地铁：施工照片审核，从人工逐张查看，到AI自动搞定95%

极视角开创了AI视觉算法商城，拥有高精度的新一代多模态视觉语言大模型以及 1500余种成熟图像识别算法，覆盖超过100个行业领域的应用场景，已成功服务过3000 多家政企与科研院所，汇聚了超过数十万海内外视觉算法开发者。

Shandong Extreme Vision pioneered the AI Vision Algorithm Marketplace. The company offers a next-generation, high-precision multimodal vision-language model and more than 1,500 proven image recognition algorithms, covering use cases across more than 100 industries. It has served over 3,000 government and enterprise clients and research institutes, and has brought together hundreds of thousands of vision algorithm developers from China and overseas.

星际视觉语言大模型（Stellaris-VL），是极视角基于十年Al技术沉淀与10亿+真实业务场景数据集，自主研发的面向产业的高精度视觉语言大模型。

The Stellaris Vision-Language Model (Stellaris-VL) is a high-precision, industry-oriented vision-language model independently developed by Extreme Vision. It is built on a decade of AI expertise and a dataset of over 1 billion real-world business scenarios.

企业负责人向山东国际传播中心记者介绍，地铁施工前后都要拍照，每天几百个点位，从前全靠人工肉眼比对。如今有了星际大模型，智能体自动判断95%，人只用复核剩下的5%，效率大大提高。

A company executive told a reporter from the Shandong International Communication Center that, for metro construction, photos are taken before and after construction work at hundreds of locations every day, and that in the past all comparisons were done manually by eye. Now, with the Stellaris model, an AI agent automatically handles 95% of the assessments, leaving people to review only the remaining 5%—greatly improving efficiency.