Person Re-Identification by Deep Joint Learning of Multi-Loss Classification

By Xiatian Zhu, Shaogang Gong. International Joint Conference on Artificial Intelligence, 2017

426 citations recorded by the source corpus.

Existing person re-identification (re-id) methods rely mostly on either localised or global feature representation. This ignores their joint benefit and mutual complementary effects. In this work, we show the advantages of jointly learning local and global features in a Convolutional Neural Network (CNN) by aiming to discover correlated local and global features in different context. Specifically, we formulate a method for joint learning of local and global feature selection losses designed to optimise person re-id when using generic matching metrics such as the L2 distance. We design a novel CNN architecture for Jointly Learning Multi-Loss (JLML) of local and global discriminative feature optimisation subject concurrently to the same re-id labelled information. Extensive comparative evaluations demonstrate the advantages of this new JLML model for person re-id over a wide range of state-of-the-art re-id methods on five benchmarks (VIPeR, GRID, CUHK01, CUHK03, Market-1501).

Subfields: Computer Vision and Pattern Recognition, Ocean Engineering

View Paper: https://www.semanticscholar.org/paper/65510920d60f23faae27e073c11e4f037c538067

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

In Ocean Engineering

Citations in the corpus

Knowledge Distillation by On-the-Fly Native Ensemble (2018, background): “Deep neural networks have gained impressive success in many computer vision tasks [1; 2; 3; 4; 5; 6; 7].”
Joint Discriminative and Generative Learning for Person Re-Identification (2019, background): “One of the common practice is to split input images or feature maps horizontally to take advantage of local spatial cues [22, 38, 48].”
Adaptive Transfer Network for Cross-Domain Person Re-Identification (2019, methodology): “[14] formulated a method of jointly learning local and global features in a CNN model by optimizing multiple classification losses in different context.”
Person Re-identification by Deep Learning Multi-scale Representations (2017, methodology): “Our DPFL outnumbers the deep local-global joint CNN model JLML [27] by 3.”
FD-GAN: Pose-guided Feature Distilling GAN for Robust Person Re-identification (2018, methodology): “We compare our proposed FD-GAN with the state-of-the-art person reID methods including VI+LSRO [3], JLML [50], PA [48], etc.”
Learning Discriminative Features with Multiple Granularities for Person Re-Identification (2018, methodology): “In [8, 21, 36], images are all split into several stripes in horizontal orientation according to intrinsic human body structure knowledge, on which local feature representations are learned.”
Transferable Joint Attribute-Identity Deep Learning for Unsupervised Person Re-identification (2018, methodology): “Most existing re-id studies follow the supervised learning paradigm such as optimising pairwise matching distance metrics [20, 52, 57, 49, 54, 47, 50, 5] or deep learning methods [27, 45, 51, 44, 4, 28].”
Pyramidal Person Re-IDentification via Multi-Loss Dynamic Training (2018, methodology): “Some methods [13, 26] aware of that integrating the local and global features can improve the performance but the information between them is also ignored.”
Human Semantic Parsing for Person Re-identification (2018, neutral): “7 JLML [25] 65.”
Horizontal Pyramid Matching for Person Re-identification (2018, background): “Third, the partition is cropped into pre-deﬁned patches (Sun et al. 2017b; Li et al. 2017).”
Beyond Part Models: Person Retrieval with Refined Part Pooling (2017, neutral): “MultiLoss [19] 83.”
Attribute-Aware Attention Model for Fine-grained Representation Learning (2018, background): “N, which is much similar to our Baseline-2. We come up with two attention models which are identity-attribute reciprocal to further improve the usage of attribute features and regional features. JLML [41] is designed to share identity label constraints by allocating each attribute branch with loss f”
RGB-Infrared Cross-Modality Person Re-Identification via Joint Pixel and Feature Alignment (2019, methodology): “Existing methods can be grouped into hand-crafted descriptors [22, 36, 19, 34], metric learning methods [43, 16, 20] and deep learning algorithms [42, 11, 28, 21, 44, 30, 15, 35, 33, 18, 17, 1, 31].”