RefineFace: Refinement Neural Network for High Performance Face Detection

TL;DR

RefineFace is a single-shot face detector that integrates multiple modules to improve accuracy and efficiency, especially for tiny faces, achieving state-of-the-art results on standard benchmarks.

Contribution

The paper introduces a novel refinement neural network with five modules designed to enhance high-performance face detection, particularly for tiny and challenging faces.

Findings

Achieves state-of-the-art results on WIDER FACE and other benchmarks.

Runs at 37.3 FPS with ResNet-18 on VGA images.

Effectively detects faces in extreme poses and scales.

Abstract

Face detection has achieved significant progress in recent years. However, high performance face detection still remains a very challenging problem, especially when there exists many tiny faces. In this paper, we present a single-shot refinement face detector namely RefineFace to achieve high performance. Specifically, it consists of five modules: Selective Two-step Regression (STR), Selective Two-step Classification (STC), Scale-aware Margin Loss (SML), Feature Supervision Module (FSM) and Receptive Field Enhancement (RFE). To enhance the regression ability for high location accuracy, STR coarsely adjusts locations and sizes of anchors from high level detection layers to provide better initialization for subsequent regressor. To improve the classification ability for high recall efficiency, STC first filters out most simple negatives from low level detection layers to reduce search space for subsequent classifier, then SML is applied to better distinguish faces from background at various scales and FSM is introduced to let the backbone learn more discriminative features for classification. Besides, RFE is presented to provide more diverse receptive field to better capture faces in some extreme poses. Extensive experiments conducted on WIDER FACE, AFW, PASCAL Face, FDDB, MAFA demonstrate that our method achieves state-of-the-art results and runs at $37.3$ FPS with ResNet-18 for VGA-resolution images.