从Mobile Vision Android(MVA)迁移至MLK的技术疑问咨询

I’ve been developing a project with Mobile Vision Android (MVA) for 4 months. Implementing features was straightforward—all I needed was to include the play service dependency and follow the official Codelab tutorials.

Google recently released the new Machine Learning Kit (MLK), which offers new capabilities and is strongly recommended by Google for developers to adopt. However, ML Kit is built on Firebase, requiring setup steps like configuring a Google Developer account and completing the official setup workflow. From my perspective, this would create a tight coupling between my project and Google's ecosystem—feel free to correct me if this assumption is wrong.

I have two core technical questions:

• Is it possible to use ML Kit without configuring Firebase, just like how MVA works (only adding dependencies)? (This question has already been answered by @Ian Barber)
• My current app uses MVA’s Codelab to implement text detection on camera video streams. MVA handles frame capture and processing with optimized multi-threading out of the box. But ML Kit doesn’t provide any video processing examples at the moment. How can I achieve camera-based text detection in ML Kit using components like CameraSource and CameraSourcePreview, similar to my MVA implementation?

内容的提问来源于stack exchange，提问作者Arnauld Alex