masksdk 1.1.1

sdk for mask

image-obfuscation-sdk

an obfuscation server contains two parts. one is inference, another is mask

this repo stands for the mask part standalone SDK

TODO

1. add resize mask technology
2. more lightweight API, such as numpy-op directly instead of PIL-op

How to

import io
from masksdk.call_infer_and_mask import mask_one_image, ModelName, MaskName

# conf
infer_host = 'your_ip:your_port'

# mask in local and infer in server
raw_img_file = 'path_to_your_image_file/image_file_name.jpg'
with open(raw_img_file, "rb") as img_content:
    mask_img, bbox = mask_one_image(
        infer_host=infer_host,
        image_file=io.BytesIO(img_content.read()),
        model_names=[ModelName.face_ai, ModelName.plate_ai],
        mask_name=MaskName.blur,
        degree=0.5,
        down_sampling_ratio=1,
        pixel_gap=10,
        api_version='v2',
    )

# show
mask_img.show()
print('bbox={}'.format(bbox))