Products

社区开发者基于 MiniCPM-V 4.7 1B 打造视障辅助应用 PRTS

Why we picked it

OpenBMB 社区有人用 1B 的 MiniCPM-V 做了个 iPhone 视障辅助应用,拍照描述环境、读文本都能离线跑。想看端侧小模型能干啥、或者做无障碍开发的话,这个案例很具体,GitHub 上模型也开源了。

社区开发者 Rebel-Gloria 发布了 iOS 应用 PRTS,面向盲人和低视力用户,默认采用 OpenBMB 的 MiniCPM-V 4.7 1B 模型在设备端进行视觉理解。

Body

应用核心功能包括:拍照生成环境场景描述、拍摄标识或印刷文本并朗读、通过语音输入针对当前画面追问细节;支持设备上另有基于 LiDAR 的避障系统,提供语音方向提示和节奏性触觉反馈。交互层面适配了 VoiceOver、Dynamic Type、实体按钮、Back Tap、Siri 快捷指令,并为低视力用户保留了相机预览的专用模式。

常规场景描述和文本朗读可完全离线运行,得益于 1B 参数的小模型可在手机端部署;在线功能和系统语音识别则需要联网。这款应用展示了开源端侧多模态模型在无障碍场景的实际落地。

Original · OpenBMB

📱 Point your camera. Hear what’s in front of you.

Community developer Rebel-Gloria built PRTS, an iPhone and iPad app for blind and low-vision users, with MiniCPM-V 4.7 1B powering its default on-device visual understanding. 💙

👀 Scene descriptions: Capture an image and hear a concise description of your surroundings.

📖 Read text: Photograph signs, notices, or printed pages and have them read aloud.

🎙️ Ask follow-up questions: Use voice input to ask about details in the current scene.

📳 Obstacle guidance: A separate LiDAR-based system provides spoken directions and rhythmic haptic cues on supported devices.

♿ Accessible controls: VoiceOver, Dynamic Type, physical buttons, Back Tap, and Siri Shortcuts offer multiple ways to interact. A dedicated low-vision mode retains the camera preview, scene descriptions, and text reading.

🔒 Routine scene descriptions and text reading can run offline with the built-in MiniCPM-V model. Optional online features and system speech recognition have separate connectivity requirements.

A thoughtful community exploration of visual intelligence for more accessible everyday experiences. 🛠️

📂 Explore PRTS: https://t.co/o0Wes81Gr1 🚀 MiniCPM-V GitHub: https://t.co/iYDxpa52tn 🤗 On-device model: https://t.co/9pKlAjgQif

#MiniCPM #OpenBMB #Accessibility #OnDeviceAI #OpenSource

Industry Coordinates & Source Verification

This report is categorized under AITOP's "Products" desk, focusing on China's AI industry and domestic foundation model ecosystem. Within our cross-regional analytical framework, it directly maps to AI developer toolchains, coding agents, and engineering workflows. Key entities and technical topics under continuous tracking include MiniCPM-V, OpenBMB, accessibility, on-device models.

Source attribution: curated from OpenBMB (@OpenBMB).