Dr. Zhiming Luo

2017-ICCV-S3FD: Single Shot Scale-invariant Face Detector

S. Zhang, et al. "S3FD: Single Shot Scale-invariant Face Detector." ICCV 2017. [pdf]

Basic Idea

A improved SSD architecture is proposed to do face detection
1. Scale-equitable framework
2. Scale compensation anchor matching strategy
3. Max-out background label

1. Scale-equitable framework

Designing scales for anchors

Effective receptive field
Equal-proportion interval principle

2. Scale compensation anchor matching strategy

A two stage

3. Max-out background label

Google Sites

Report abuse