面壁 MiniCPM 打造 AI 失物招领应用

OpenBMB · @OpenBMB · X·2026-08-21 22:00·13天前
AI 导读

开发者 @JacobLinCool 用面壁智能 MiniCPM 系列模型构建了 Lost & Found Desk 失物招领应用,将照片识别与多语言匹配结合。其中 MiniCPM-V 4.6 负责从物品照片生成可搜索描述,MiniCPM5-1B 则理解用户多语言描述并匹配失物记录,展示了小模型在真实场景中的应用潜力。

OpenBMB@OpenBMB
29AI 编辑部评分,满分 100

面壁 MiniCPM 打造 AI 失物招领应用

2026-08-21 22:00· 13天前
AI 导读

开发者 @JacobLinCool 用面壁智能 MiniCPM 系列模型构建了 Lost & Found Desk 失物招领应用,将照片识别与多语言匹配结合。其中 MiniCPM-V 4.6 负责从物品照片生成可搜索描述,MiniCPM5-1B 则理解用户多语言描述并匹配失物记录,展示了小模型在真实场景中的应用潜力。

What can you build with small multimodal models beyond demos?

Lost & Found Desk built by developer @JacobLinCool turns MiniCPM into an AI-powered lost & found workflow — from understanding item photos to helping people describe, match, and recover what they lost. 🔍

Powered by MiniCPM-V 4.6 and MiniCPM5-1B, this project enables:

📸 MiniCPM-V 4.6 to generate searchable descriptions from photos of found items taken by staff. 🌍 MiniCPM5-1B to understand multilingual natural language descriptions from users and match them with lost item records.

By combining visual understanding and language reasoning, MiniCPM helps make lost-and-found services smarter, faster, and more accessible.

🔗 Project: https://huggingface.co/spaces/build-small-hackathon/lost-found-desk

🤗 Models: MiniCPM-V 4.6: https://huggingface.co/openbmb/MiniCPM-V-4.6 MiniCPM5-1B: https://huggingface.co/openbmb/MiniCPM5-1B