RetinaFace: Single-shot Multi-level Face Localization in the Wild
RetinaFace: Single-shot Multi-level Face Localization in the Wild
复制标题
DOI：
--
复制
发表时间：
2021
期刊：
影响因子：
--
通讯作者：
中科院分区：
文献类型：
--
作者：
RetinaFace[2]是一个深度学习模型，它通过为每个单独的人脸提出矩形区域（边界框）3来检测图像中的人脸。与其他当前最先进的模型不同，本研究提出了一种多任务损失计算方法，同时计算5个面部标志（眼睛、鼻子和嘴巴两侧）的坐标，并同时计算1000个点的3D面部网格。此外，该模型还采用了级联结构[13]和可变形卷积层（DCL）[1]。本文的研究范围包括不包括DCL的整个模型结构。此外，由于3D 8点检测数据库不是公开共享的，因此所实现的任务仅限于人脸边界框检测和地标定位任务。9
RetinaFace [2] is a deep learning model that detects faces in images by proposing rectangular areas (bounding boxes) 3 for every single face. Unlike the other current state-of-the-art models, this study proposes a multi-task loss calculation 4 by also computing the coordinates of 5 facial landmarks (eyes, nose, and two sides of the mouth) and 3D face mesh 5 with 1000 points concurrently. Additionally, the proposed model also adapts a cascaded structure [13] and deformable 6 convolution layers (DCL) [1]. The scope of this paper includes the whole model structure excluding DCL. Additionally, 7 The tasks implemented are limited only to face bounding box detection and landmark localization tasks, since the 3D 8 point detection database is not publicly shared. 9