Deep feature learning with relative distance comparison for person re-identification

By Shengyong Ding, Liang Lin, Guangrun Wang, Hongyang Chao. 2015

686 citations recorded by the source corpus.

Identifying the same individual across different scenes is an important yet difficult task in intelligent video surveillance. Its main difficulty lies in how to preserve similarity of the same person against large appearance and structure variation while discriminating different individuals. In this paper, we present a scalable distance driven feature learning framework based on the deep neural network for person re-identification, and demonstrate its effectiveness to handle the existing challenges. Specifically, given the training images with the class labels (person IDs), we first produce a large number of triplet units, each of which contains three images, i.e. one person with a matched reference and a mismatched reference. Treating the units as the input, we build the convolutional neural network to generate the layered representations, and follow with the L 2 distance metric. By means of parameter optimization, our framework tends to maximize the relative distance between the matched pair and the mismatched pair for each triplet unit. Moreover, a nontrivial issue arising with the framework is that the triplet organization cubically enlarges the number of training triplets, as one image can be involved into several triplet units. To overcome this problem, we develop an effective triplet generation scheme and an optimized gradient descent algorithm, making the computational load mainly depend on the number of original images instead of the number of triplets. On several challenging databases, our approach achieves very promising results and outperforms other state-of-the-art approaches. HighlightsWe present a novel feature learning framework for person re-identification.Our framework is based on the maximum relative distance comparison.The learning algorithm is scalable to process large amount of data.We demonstrate superior performances over other state-of-the-arts.

Subfields: Biomedical Engineering, Computer Vision and Pattern Recognition

How researchers see this paper

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

In Biomedical Engineering

1. Liang Zheng
2. Yi Yang
3. Qi Tian
4. Zhedong Zheng
5. Shiliang Zhang
6. Shaogang Gong
7. Tao A. Xiang
8. Wanli Ouyang

In Computer Vision and Pattern Recognition

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

Citations in the corpus

• Recent advances in convolutional neural networks (2015, methodology): “In this section, we introduce some fast processing methods of CNNs.”
• In Defense of the Triplet Loss for Person Re-Identification (2017, methodology): “Furthermore we also show that a margin-less formulation performs slightly better, while removing one hyperparameter.”
• MARS: A Video Benchmark for Large-Scale Person Re-Identification (2016, methodology): “Image pairs [1,23,40] or triplets [8] are defined as input to CNN, instead of single training images.”
• Real-World Anomaly Detection in Surveillance Videos (2018, methodology): “have been used for feature learning [38], highlight detection [41], Graphics Interchange Format (GIF) generation [17], face detection and verification [33], person re-identification [13], place recognition [6], metric learning and image retrieval [16].”
• Reinforced Temporal Attention and Split-Rate Transfer for Depth-Based Person Re-identification (2017, background): “As such, there is vast literature in learning discriminative distance metrics [35,42,105,51,55,77,43,61,53,19,48,108,91,5] and discriminant subspaces [63,47,107,84,43,46,64,94,15].”
• Person Re-identification by Deep Learning Multi-scale Representations (2017, background): “Recently, deep learning re-id models [26, 4, 48, 56, 44, 77, 64, 7, 11, 65, 9, 60, 27] start to take over and have obtained impressive performance.”
• Learning Deep Context-Aware Features over Body and Latent Parts for Person Re-identification (2017, background): “In addition to the deep metric learning, some work directly learns image representation through pair-wise contrastive loss or triplet ranking loss, and use Euclidean metric for comparison [5, 6, 34, 35].”
• End-to-End Deep Learning for Person Search (2016, methodology): “[5] utilized triplet samples for training CNNs to minimize the feature distance between the same person, while maximize the distance between different people.”
• Person Recognition System Based on a Combination of Body Images from Visible Light and Thermal Cameras (2017, methodology): “- Higher identification accuracy can be obtained; the extracted image features are slightly invariant to noise, illumination conditions, and misalignment because of the use of deep learning method [14,15,25].”
• Learning Deep Neural Networks for Vehicle Re-ID with Visual-spatio-Temporal Path Proposals (2017, neutral): “[8] and Cheng et al.”
• Human Semantic Parsing for Person Re-identification (2018, positive): “Others [12, 9, 11] have employed triplet loss which requires a tuple of anchor, positive and negative images where the training objective is to simultaneously pushing the positive image towards the anchor while pulling the negative image away from it.”
• See the Forest for the Trees: Joint Spatial and Temporal Recurrent Neural Networks for Video-Based Person Re-identification (2017, neutral): “And more importantly, deep neural networks have shown to be effective for person re-identification, and have achieved much better performance than traditional methods [1, 6, 9, 12, 13, 18, 25, 36, 38].”
• Deep Growing Learning (2017, background): “Deep convolutional neural networks (CNNs) [22, 21, 12] have exhibited state-of-the-art performance in image classification [22, 21, 35, 4], and the prediction accuracy of the CNN-based classifier can be improved by training on massive amounts of data (e.”
• Attention-Aware Compositional Network for Person Re-identification (2018, background): “The second category [28, 12, 27, 45, 19, 9, 23, 14, 33] mainly targets on learning a robust distance metric to measure the similarity between images.”
• Embedding Deep Metric for Person Re-identification: A Study Against Large Variations (2016, methodology): “The re-identification methods by deep learning [1,6,17,35] incorporate the two above-mentioned aspects (feature extraction and metric learning) of person re-identification into an integrated framework.”
• Person Re-identification in the Wild (2016, methodology): “The former calculate the average precision corresponding to ten recalls uniformly sampled from [0, 1] [15], while MR is the average miss rate at 9 False Positive Per Image (FPPI) uniformly sampled from [10−2, 100] [10].”
• Joint Learning of Single-Image and Cross-Image Representations for Person Re-identification (2016, methodology): “proposed a deep SIR learning model based on relative distance comparison for person re−identification [7].”