Face-MagNet: Magnifying Feature Maps to Detect Small Faces

TL;DR

Face-MagNet introduces a novel face detection architecture that enhances small face detection by magnifying feature maps through deconvolution layers, outperforming some existing methods on standard datasets.

Contribution

The paper proposes Face-MagNet, a new face detector that improves small face detection without skip or residual connections, using deconvolution layers within the Faster-RCNN framework.

Findings

Face-MagNet outperforms ResNet101-based HR on WIDER dataset.

Achieves comparable results to state-of-the-art methods like SSH.

Effective in detecting small faces with enhanced feature map resolution.

Abstract

In this paper, we introduce the Face Magnifier Network (Face-MageNet), a face detector based on the Faster-RCNN framework which enables the flow of discriminative information of small scale faces to the classifier without any skip or residual connections. To achieve this, Face-MagNet deploys a set of ConvTranspose, also known as deconvolution, layers in the Region Proposal Network (RPN) and another set before the Region of Interest (RoI) pooling layer to facilitate detection of finer faces. In addition, we also design, train, and evaluate three other well-tuned architectures that represent the conventional solutions to the scale problem: context pooling, skip connections, and scale partitioning. Each of these three networks achieves comparable results to the state-of-the-art face detectors. With extensive experiments, we show that Face-MagNet based on a VGG16 architecture achieves better results than the recently proposed ResNet101-based HR method on the task of face detection on WIDER dataset and also achieves similar results on the hard set as our other method SSH.