Multi-Path Region-Based Convolutional Neural Network for Accurate Detection of Unconstrained "Hard Faces"
This work addresses the problem of detecting faces across extreme size variations in unconstrained images, which is crucial for applications like surveillance and social media, but it appears incremental as it builds on existing region-based CNN methods.
The paper tackled the challenge of large-scale variations in unconstrained face detection by proposing a two-stage cascaded framework, MP-RCNN, which achieved state-of-the-art performance on the WIDER FACE dataset 'hard' partition with a 9.6% improvement in Average Precision.
Large-scale variations still pose a challenge in unconstrained face detection. To the best of our knowledge, no current face detection algorithm can detect a face as large as 800 x 800 pixels while simultaneously detecting another one as small as 8 x 8 pixels within a single image with equally high accuracy. We propose a two-stage cascaded face detection framework, Multi-Path Region-based Convolutional Neural Network (MP-RCNN), that seamlessly combines a deep neural network with a classic learning strategy, to tackle this challenge. The first stage is a Multi-Path Region Proposal Network (MP-RPN) that proposes faces at three different scales. It simultaneously utilizes three parallel outputs of the convolutional feature maps to predict multi-scale candidate face regions. The "atrous" convolution trick (convolution with up-sampled filters) and a newly proposed sampling layer for "hard" examples are embedded in MP-RPN to further boost its performance. The second stage is a Boosted Forests classifier, which utilizes deep facial features pooled from inside the candidate face regions as well as deep contextual features pooled from a larger region surrounding the candidate face regions. This step is included to further remove hard negative samples. Experiments show that this approach achieves state-of-the-art face detection performance on the WIDER FACE dataset "hard" partition, outperforming the former best result by 9.6% for the Average Precision.