Yes of course, Qwen2-VL has capabilities of understanding custom documents and retrieve data in almost real time depending upload the size of the image. You can test out the base model or else try to fine-tune the model. For any help, connect with me in LinkedIn