首页> 外国专利> HUMAN-FACE DETECTION MODEL TRAINING METHOD, DEVICE, MEDIUM, AND HUMAN-FACE DETECTION METHOD

HUMAN-FACE DETECTION MODEL TRAINING METHOD, DEVICE, MEDIUM, AND HUMAN-FACE DETECTION METHOD

机译:人脸检测模型训练方法,装置,培养基和人脸检测方法

摘要

Provided are a human-face detection model training method, device, computer-readable storage medium, and human-face detection method, said human-face detection model training method comprising: extracting features from an image to be trained to obtain initial feature maps of at least two resolutions (S10); fusing said initial feature maps to obtain a first detection feature map and a second detection feature map (S20); performing first-scale face shape detection on the first detection feature map by means of a first detector, and performing second-scale face shape detection on second detection feature map by means of a second detector (S30); training the human-face detection model on the basis of a target loss function. The described human-face detection model training method has the advantage of improving the detection rate of human faces of various sizes.
机译:提供了人面检测模型训练方法,装置,计算机可读存储介质和人脸检测方法,所述人脸检测模型训练方法包括:从要训练的图像中提取特征以获得初始特征图 至少有两项决议(S10); 融合所述初始特征映射以获得第一检测特征图和第二检测特征映射(S20); 通过第一检测器执行在第一检测特征图上对第一检测特征映射进行第一尺度面部形状检测,并通过第二检测器对第二检测特征映射执行第二尺度面部形状检测(S30); 基于目标损失功能训练人脸检测模型。 所描述的人脸检测模型训练方法具有提高各种尺寸的人面的检测率的优点。

著录项

相似文献

  • 专利
  • 外文文献
  • 中文文献
获取专利

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号