初始化 Pix2Text Args: device: 计算设备(cuda 使用 GPU3) languages: 文字语言(影响非公式部分的识别)
(self, device: str = 'cuda', languages: tuple = ('en',))
| 70 | """ |
| 71 | |
| 72 | def __init__(self, device: str = 'cuda', languages: tuple = ('en',)): |
| 73 | """ |
| 74 | 初始化 Pix2Text |
| 75 | |
| 76 | Args: |
| 77 | device: 计算设备(cuda 使用 GPU3) |
| 78 | languages: 文字语言(影响非公式部分的识别) |
| 79 | """ |
| 80 | print(f" Pix2Text 使用设备: {device}") |
| 81 | |
| 82 | # 降低 MFD 检测阈值,提高公式检测率 |
| 83 | self.p2t = Pix2Text.from_config( |
| 84 | device=device, |
| 85 | text_config={'device': device, 'languages': languages}, |
| 86 | formula_config={'device': device}, |
| 87 | # MFD 配置:降低置信度阈值 |
| 88 | mfd_config={ |
| 89 | 'device': device, |
| 90 | 'conf_threshold': 0.15, # 默认 0.25,降低以检测更多公式 |
| 91 | 'iou_threshold': 0.45, # 默认 0.45 |
| 92 | }, |
| 93 | ) |
| 94 | |
| 95 | def analyze_image(self, image_path: str) -> Pix2TextResult: |
| 96 | """ |
nothing calls this directly
no outgoing calls
no test coverage detected