Quality Aware Network for Set to Set Recognition

By Yu Liu, Junjie Yan, Wanli Ouyang. Computer Vision and Pattern Recognition, 2017

327 citations recorded by the source corpus.

This paper targets on the problem of set to set recognition, which learns the metric between two image sets. Images in each set belong to the same identity. Since images in a set can be complementary, they hopefully lead to higher accuracy in practical applications. However, the quality of each sample cannot be guaranteed, and samples with poor quality will hurt the metric. In this paper, the quality aware network (QAN) is proposed to confront this problem, where the quality of each sample can be automatically learned although such information is not explicitly provided in the training stage. The network has two branches, where the first branch extracts appearance feature embedding for each sample and the other branch predicts quality score for each sample. Features and quality scores of all samples in a set are then aggregated to generate the final feature embedding. We show that the two branches can be trained in an end-to-end manner given only the set-level identity annotation. Analysis on gradient spread of this mechanism indicates that the quality learned by the network is beneficial to set-to-set recognition and simplifies the distribution that the network needs to fit. Experiments on both face verification and person re-identification show advantages of the proposed QAN. The source code and network structure can be downloaded at GitHub.

Subfields: Computer Vision and Pattern Recognition, Signal Processing

View Paper: https://www.semanticscholar.org/paper/08d55271589f989d90a7edce3345f78f2468a7e0

How researchers see this paper

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

In Computer Vision and Pattern Recognition

1. Yi Yang
2. Liang Zheng
3. Xiaogang Wang
4. Luc Van Gool
5. Wanli Ouyang
6. Dacheng Tao
7. G. Wang
8. Ming-Hsuan Yang

In Signal Processing

1. Xuanyi Dong
2. Yi Yang
3. Liang Zheng
4. Wanli Ouyang
5. Hang Zhou
6. Yu Liu
7. Mang Ye
8. Pong Chi Yuen

Citations in the corpus

• Part-Aligned Bilinear Representations for Person Re-identification (2018, unknown): “] (Video) 68:3 82:6 - 89:4 49:3 Liu et al. [86] (Video) 68:3 81:4 - 90:6 52:9 Zhou et al. [87] 70:6 90:0 - 97:6 50:7 Li et al. [17] 71:8 86:6 - 93:1 56:1 + re-ranking 83:0 93:7 - 97:6 66:4 Liu et al. [88] 73:7 84:9 - 91:6 51:7 Hermans et al. [72] 79:8 91:4 - - 67:7 + re-ranking 81:2 90:8 - - 77:4 Pr”
• Deep Cosine Metric Learning for Person Re-identification (2018, unknown): “06 P-QAN [17] 73.”
• Robust Anchor Embedding for Unsupervised Video Person re-IDentification in the Wild (2018, methodology): “In addition, some CNN-RNN related network structures are also designed for video re-ID task [26, 30, 44].”
• A Pose-Sensitive Embedding for Person Re-identification with Expanded Cross Neighborhood Re-ranking (2017, unknown): “ - - - - PDC [37] ICCV17 63.4 84.1 - - - - JLML [23] IJCAI17 65.5 85.1 - - - - DPFL [5] ICCV17-W 72.6 88.6 60.6 79.2 - - Forest [60] CVPR17 - - - - 50.7 70.6 DGM+IDE [44] ICCV17 - - - - 46.8 65.2 QMAy[26] CVPR17 - - - - 51.7 73.7 Our ResNet-50 Baseline 59.8 82.6 50.3 71.5 49.5 64.5 PSE 69.0 87.7 62.”
• Discriminability Distillation in Group Representation Learning (2020, methodology): “[37] propose the Quality Aware Network (QAN) to learn a quality score for each image inside an image set during network training.”
• Dual Attention Matching Network for Context-Aware Feature Sequence Based Person Re-identification (2018, methodology): “For example, in [33, 26], high-quality local patterns are explored from images or videos firstly and then aggregated into informative feature vectors; in [28, 39, 49], local features of recurrent appearance data are extracted and integrated using temporal-pooling strategy; in [14, 5], to enhance the”
• Towards Real-Time Multi-Object Tracking (2019, methodology): “In comparison, Faster R-CNN + QAN embedding only runs at 6 FPS with MOTA=66.1% on the MOT-16 test set.”
• Exploit the Unknown Gradually: One-Shot Video-Based Person Re-identification by Stepwise Learning (2018, methodology): “In recent years, person re-ID on the large-scale video data, such as surveillance videos, has attracted significant attention [10, 20, 28, 32, 35].”
• Counterfactual Attention Learning for Fine-Grained Visual Categorization and Re-identification (2021, unknown): “Another group of methods [32, 48, 26, 8] adopts attention mechanism on video-based person re-identification task to discover key parts in videos.”
• Search to Distill: Pearls Are Everywhere but Not the Eyes (2019, background): “Over the past few years, Deep Neural Networks (DNNs) have achieved unprecedented success on various tasks, such as image understanding [16, 24], image generation [23], machine translation [37] and speech recognition [7].”