本地部署教程

[win4r]

https://youtu.be/IJsDfoZg2QM

[abrichr]

请参考 https://github.com/microsoft/OmniParser/pull/52 中的Dockerfile，该文件已实现。使用方法如下：

% sudo nvidia-docker build -t omniparser .
% sudo docker run -d -p 7861:7861 --gpus all --name omniparser-container omniparser
% python client.py "http://0.0.0.0:7861" ~/Desktop/screenshot.png
Loaded as API: http://44.198.58.162:7861/ ✔                                                                                                                                                                                                               
2024-10-29 18:55:51.336 | INFO     | __main__:predict:61 - Prediction completed successfully                                                                                                                                                              
2024-10-29 18:55:51.337 | INFO     | __main__:predict:70 - label_coordinates={'0': [0.12276943773031235, 0.002296212362125516, 0.045681655406951904, 0.028702640905976295],
...                 
2024-10-29 18:55:51.339 | INFO     | __main__:predict:71 - parsed_content_list=['Text Box ID 0: 5 v G', 'Text Box ID 1: Book1', 'Text Box ID 2: Home', 
...
2024-10-29 18:55:51.341 | INFO     | __main__:predict:77 - Parsed content saved to: result_data_20241029_185545.json         
2024-10-29 18:55:51.343 | INFO     | __main__:predict:83 - Output image saved to: output_image_20241029_185545.png  
% open output_image.png 
% head result_data_20241029_185545.json                                              
{                                                                                                                            
    "label_coordinates": {                                                                                                   
        "0": [                                                                                                               
            0.12276943773031235,                                                                                             
            0.002296212362125516,                                                                                            
            0.045681655406951904,                                                                                            
            0.028702640905976295                                                                                             
        ],                                                                                                                   
        "1": [                                                                                                               
            0.48429691791534424,                                                                                             
% tail result_data_20241029_185545.json                                              
        "Icon Box ID 72: Decrease",                                                                                          
        "Icon Box ID 73: Download",                                                                                          
        "Icon Box ID 74: a menu or list.",                                                                                   
        "Icon Box ID 75: navigating to the previous item or screen.",                                                        
        "Icon Box ID 76: the \"E\" function.",                                                                               
        "Icon Box ID 77: Justified",                                                                                         
        "Icon Box ID 78: Text formatting options.",                                                                          
        "Icon Box ID 79: a search function."                                                                                 
    ]                                                                                                                        
}%