wehub-resource-sync
39a086b41b
Build Windows CPU / build (push) Has been cancelled
Build Windows CUDA 10.2 / build (push) Has been cancelled
Build Windows CUDA 11.8 / build (push) Has been cancelled
Build Windows CUDA 12.6 / build (push) Has been cancelled
Build Windows DirectML / build (push) Has been cancelled
38 lines
1.2 KiB
Python
38 lines
1.2 KiB
Python
|
||
from backend.tools.paddle_model_config import PaddleModelConfig
|
||
from backend.tools.hardware_accelerator import HardwareAccelerator
|
||
import numpy as np
|
||
|
||
try:
|
||
from paddleocr import TextDetection
|
||
except ImportError:
|
||
TextDetection = None
|
||
|
||
|
||
class SubtitleDetect:
|
||
"""
|
||
文本框检测类，用于检测视频帧中是否存在文本框
|
||
"""
|
||
|
||
def __init__(self):
|
||
hardware_accelerator = HardwareAccelerator.instance()
|
||
model_config = PaddleModelConfig(hardware_accelerator)
|
||
# 使用 TextDetection 公开 API（PaddleOCR 3.x）
|
||
kwargs = {'model_dir': model_config.DET_MODEL_PATH}
|
||
if model_config.DET_MODEL_NAME:
|
||
kwargs['model_name'] = model_config.DET_MODEL_NAME
|
||
self.text_detector = TextDetection(**kwargs)
|
||
|
||
def detect_subtitle(self, img):
|
||
"""
|
||
检测图像中的文本框
|
||
:param img: 输入图像
|
||
:return: (dt_boxes, elapse) dt_boxes为numpy数组，elapse为耗时
|
||
"""
|
||
results = list(self.text_detector.predict(img))
|
||
if not results:
|
||
return np.array([]), 0
|
||
res = results[0]
|
||
dt_polys = res.get('dt_polys', np.array([]))
|
||
return dt_polys, 0
|