Improving Face Detection Performance with 3D-Rendered Synthetic Data

Abstract

In this paper, we provide a synthetic data generator methodology with fully controlled, multifaceted variations based on a new 3D face dataset (3DU-Face). We customized synthetic datasets to address specific types of variations (scale, pose, occlusion, blur, etc.), and systematically investigate the influence of different variations on face detection performances. We examine whether and how these factors contribute to better face detection performances. We validate our synthetic data augmentation for different face detectors (Faster RCNN, SSH and HR) on various face datasets (MAFA, UFDD and Wider Face).

arXiv comment11 pages. Submitted to Pattern Recognition Letters

References in corpus (12)

• Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks
• Face Attention Network: An Effective Face Detector for the Occluded Faces
• Face Detection through Scale-Friendly Deep Convolutional Networks
• Do We Really Need to Collect Millions of Faces for Effective Face Recognition?
• Training Deep Face Recognition Systems with Synthetic Data
• PyramidBox: A Context-assisted Single Shot Face Detector
• S${}^3$FD: Single Shot Scale-invariant Face Detector
• SFace: An Efficient Network for Face Detection in Large Scale Variations
• Seeing Small Faces from Robust Anchor's Perspective
• Recent Progress of Face Image Synthesis
• Adversarial Occlusion-aware Face Detection
• Survey of Face Detection on Low-quality Images