Finding Tiny Faces in the Wild with Generative Adversarial Network

By Yancheng Bai, Yongqiang Zhang, Ming Ding, Bernard Ghanem. 2018

210 citations recorded by the source corpus.

Face detection techniques have been developed for decades, and one of remaining open challenges is detecting small faces in unconstrained conditions. The reason is that tiny faces are often lacking detailed information and blurring. In this paper, we proposed an algorithm to directly generate a clear high-resolution face from a blurry small one by adopting a generative adversarial network (GAN). Toward this end, the basic GAN formulation achieves it by super-resolving and refining sequentially (e.g. SR-GAN and cycle-GAN). However, we design a novel network to address the problem of super-resolving and refining jointly. We also introduce new training losses to guide the generator network to recover fine details and to promote the discriminator network to distinguish real vs. fake and face vs. non-face simultaneously. Extensive experiments on the challenging dataset WIDER FACE demonstrate the effectiveness of our proposed method in restoring a clear high-resolution face from a blurry small one, and show that the detection performance outperforms other state-of-the-art methods.

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Citations in the corpus

Bridging Non Co-occurrence with Unlabeled In-the-wild Data for Incremental Object Detection (2021, neutral): “In the task of image-based object detection, object detectors are trained on large-scale datasets and then deployed on real-world applications [9, 11, 8, 24, 5, 15, 23, 17, 16, 1, 2, 39, 33, 34, 38, 36, 37, 35, 32].”
Detecting small faces in the wild based on generative adversarial network and contextual information (2019, methodology): “This paper is an extended version of our CVPR18 oral paper [12] .”
Multi-task Generative Adversarial Network for Detecting Small Objects in the Wild (2020, methodology): “2, we adopt a deep CNN architecture which has shown effectiveness for image de-blurring in (Hradiš et al. 2015), general object detection in (Bai et al. 2018b) and face detection in (Bai et al. 2018a).”
C2AE: Class Conditioned Auto-Encoder for Open-Set Recognition (2019, methodology): “Recent advancements in computer vision have resulted in significant improvements for tasks such as image classification [16], [24], [17], [48], Detection [41], [40], [4], [49], Clustering [19], [1], [8], [2], etc.”
Towards Large-Scale Small Object Detection: Survey and Benchmarks (2022, methodology): “Bai et al. [117] extended this paradigm to face detection task and Na et al. [118] applied super-resolution method to small candidate regions for better performance.”
KGSNet: Key-Point-Guided Super-Resolution Network for Pedestrian Detection in the Wild (2020, background): “2, the backbone of the super-resolution network is a deep CNN architecture, which has shown effectiveness for image deblur-ring [65], tiny face detection [66], [67], and small general object detection [68].”
SOD-MTGAN: Small Object Detection via Multi-Task Generative Adversarial Network (2018, methodology): “As shown in Table 1 and Figure 2, we adopt a deep CNN architecture which has shown effectiveness for image de-blurring in [13] and face detection in [1].”
A Survey of the Four Pillars for Small Object Detection: Multiscale Representation, Contextual Information, Super-Resolution, and Region Proposal (2022, background): “The common overﬂow of GAN-based methods is exhibited in Fig.”
Slicing Aided Hyper Inference and Fine-Tuning for Small Object Detection (2022, background): “[21] proposed an algorithm to directly generate a clear high-resolution face from a blurry small one by adopting a generative adversarial network (GAN).”
Extended Feature Pyramid Network for Small Object Detection (2020, background): “Image-level SR is adopted in some speciﬁc situations where extremely small objects exist, such as satellite images [28] and images with crowded tiny faces [29].”
MEAL: Multi-Model Ensemble via Adversarial Learning (2018, methodology): “This procedure improves the robustness of training for student network and has applied to many ﬁelds such as image generation (Johnson, Gupta, and Fei-Fei 2018), detection (Bai et al. 2018), etc. Knowledge Transfer.”