model_watermark_detect/watermark_verify/verify_tool_mix.py

from watermark_verify.exceptions import BusinessException
from watermark_verify.process import (
    classification_all_whitebox_process,
    classification_pytorch_blackbox_process,
    classification_tensorflow_blackbox_process,
    googlenet_all_whitebox_process,
    faster_rcnn_pytorch_blackbox_process,
    faster_rcnn_pytorch_whitebox_process,
    ssd_pytorch_blackbox_process,
    ssd_pytorch_whitebox_process,
    yolox_pytorch_blackbox_process,
    yolox_pytorch_whitebox_process,
    vggnet_whitebox_process,
)

def label_verification(model_filename: str, framework: str='pytorch', mode: str='blackbox', model_type: str='yolox') -> bool:
    """
    模型标签提取验证
    :param model_filename: 模型权重文件（onnx格式）
    :param framework: 框架类型 ('pytorch' 或 'tensorflow')
    :param mode: 验证模式 ('blackbox' 或 'whitebox')
    :param model_type: 模型类型，例如 'alexnet', 'vggnet', 'googlenet', 'resnet',
                       或 'fasterrcnn', 'ssd', 'yolox'
    :return: 模型标签验证结果
    """
    model_type = model_type.lower()
    framework = framework.lower()
    mode = mode.lower()
    
    try:
        processor_class = None

        # 分类模型处理逻辑
        if model_type in ['alexnet', 'vggnet', 'googlenet', 'resnet']:
            if mode == 'blackbox':
                if framework == 'pytorch':
                    processor_class = classification_pytorch_blackbox_process.ModelWatermarkProcessor(model_filename)
                elif framework == 'tensorflow':
                    processor_class = classification_tensorflow_blackbox_process.ModelWatermarkProcessor(model_filename)
            elif mode == 'whitebox':
                if mode == 'whitebox' and model_type == 'googlenet':
                    processor_class = googlenet_all_whitebox_process.ModelWatermarkProcessor(model_filename)
                elif mode == 'whitebox' and model_type == 'vggnet':
                    processor_class = vggnet_whitebox_process.ModelWatermarkProcessor(model_filename)
                else:
                    processor_class = classification_all_whitebox_process.ModelWatermarkProcessor(model_filename)

        # 目标检测模型处理逻辑
        elif model_type == 'fasterrcnn':
            if framework == 'pytorch' and mode == 'blackbox':
                processor_class = faster_rcnn_pytorch_blackbox_process.ModelWatermarkProcessor(model_filename)
            elif framework == 'pytorch' and mode == 'whitebox':
                processor_class = faster_rcnn_pytorch_whitebox_process.ModelWatermarkProcessor(model_filename)

        elif model_type == 'ssd':
            if framework == 'pytorch' and mode == 'blackbox':
                processor_class = ssd_pytorch_blackbox_process.ModelWatermarkProcessor(model_filename)
            elif framework == 'pytorch' and mode == 'whitebox':
                processor_class = ssd_pytorch_whitebox_process.ModelWatermarkProcessor(model_filename)

        elif model_type == 'yolox':
            if framework == 'pytorch' and mode == 'blackbox':
                processor_class = yolox_pytorch_blackbox_process.ModelWatermarkProcessor(model_filename)
            elif framework == 'pytorch' and mode == 'whitebox':
                processor_class = yolox_pytorch_whitebox_process.ModelWatermarkProcessor(model_filename)

        if processor_class is None:
            raise BusinessException(
                code=-2,
                message=f"不支持的组合: framework={framework}, mode={mode}, model_type={model_type}"
            )
        if mode == 'whitebox' and model_type == 'vggnet':
            result = processor_class.process((3, 7))
        else:
            result = processor_class.process()

        return result

    except Exception as e:
        raise BusinessException(code=-1, message=str(e))

    return result