支持本地Ollama Vision模型
Note : We recommend using the latest qwen2.5vl model, as it delivers excellent results 🥰👍
推荐使用minicpm-v或llava-llama3模型,以minicpm-v为佳
ollama run llava-llama3
ollama run minicpm-v
自定义Prompt(推荐):The extracted text does not need to analyze its title and body, ignore the corresponding icon, do not guess the meaning of the text, and only output the corresponding recognized content
由于本地OCR模型识别能力有限,图片识别截图时,请不要过长或过宽,以免发生传输问题,后续可能会修复,请谅解🥺
如有问题或建议,欢迎在GitHub提交issue反馈