• Corpus ID: 198967939

Fairest of Them All: Establishing a Strong Baseline for Cross-Domain Person ReID

@article{Kumar2019FairestOT,
  title={Fairest of Them All: Establishing a Strong Baseline for Cross-Domain Person ReID},
  author={Devinder Kumar and Parthipan Siva and Paul Marchwica and Alexander Wong},
  journal={ArXiv},
  year={2019},
  volume={abs/1907.12016}
}
Person re-identification (ReID) remains a very difficult challenge in computer vision, and critical for large-scale video surveillance scenarios where an individual could appear in different camera views at different times. [...] Key Method Furthermore, using lessons learned from the state-of-the-art supervised person re-identification methods, we establish a strong baseline method for cross-domain person ReID. Experiments show that a source domain composed of two of the largest person ReID domains (SYSU and…Expand
Style Normalization and Restitution for Generalizable Person Re-Identification
TLDR
The aim of this paper is to design a generalizable person ReID framework which trains a model on source domains yet is able to generalize/perform well on target domains, and to enforce a dual causal loss constraint in SNR to encourage the separation of identity-relevant features and identity-irrelevant features.
Learning to Generalize Unseen Domains via Memory-based Multi-Source Meta-Learning for Person Re-Identification
TLDR
The Memory-based Multi-Source Meta-Learning (M3L) framework to train a generalizable model for unseen domains is proposed and a meta-learning strategy is introduced to simulate the train-test process of domain generalization for learning more generalizable models.
Human Parsing with Discriminant Feature Learning for Person Re-identification
TLDR
This work designs a multi-branch deep network model to distinguish pedestrian key features from different levels for ReID, and verifies that human parsing with discriminant feature learning contributes to the performance of ReID.
Supplementary Material: Style Normalization and Restitution for Generalizable Person Re-identification
Network Details. We use ResNet-50 [9, 1, 43, 25] as our base network for both baselines and our schemes. We build a strong baseline Baseline with some commonly used tricks integrated. Similar to [1,
Efficient Rank-Based Diffusion Process with Assured Convergence
TLDR
An efficient Rank-based Diffusion Process which combines both approaches and avoids the drawbacks of each one is proposed, capable of efficiently approximating a diffusion process by exploiting rank-based information, while assuring its convergence.
D 3 . 1 Initial Outcomes of New Learning Paradigms
This document presents the initial outcomes of the research on new learning paradigms in WP3 of AI4Media. As such, the document summarizes the research advances of the contributing partners in tasks
Building Computationally Efficient and Well-Generalizing Person Re-Identification Models with Metric Learning
TLDR
This work uses recently proposed Omni-Scale Network (OSNet) architecture combined with several training tricks and architecture adjustments to obtain state-of-the art results in cross-domain generalization problem on a large-scale MSMT17 dataset in three setups.

References

SHOWING 1-10 OF 48 REFERENCES
EANet: Enhancing Alignment for Cross-Domain Person Re-identification
TLDR
This paper proposes Part Aligned Pooling (PAP) that brings significant improvement for cross-domain testing and designs a Part Segmentation (PS) constraint over ReID feature to enhance alignment and improve model generalization and shows that applying the PS constraint to unlabeled target domain images serves as effective domain adaptation.
Person Transfer GAN to Bridge Domain Gap for Person Re-identification
TLDR
A Person Transfer Generative Adversarial Network (PTGAN) is proposed to relieve the expensive costs of annotating new training samples and comprehensive experiments show that the domain gap could be substantially narrowed-down by the PTGAN.
Generalizing a Person Retrieval Model Hetero- and Homogeneously
TLDR
A Hetero-Homogeneous Learning (HHL) method is introduced that enforces two properties simultaneously: camera invariance, learned via positive pairs formed by unlabeled target images and their camera style transferred counterparts and domain connectedness, implemented by homogeneous learning.
Generalizable Person Re-Identification by Domain-Invariant Mapping Network
TLDR
DIMN is designed to learn a mapping between a person image and its identity classifier, i.e., it produces a classifier using a single shot and can be used to match an arbitrary number of identities in a target domain.
Transferable Joint Attribute-Identity Deep Learning for Unsupervised Person Re-identification
TLDR
This work introduces an Transferable Joint Attribute-Identity Deep Learning (TJ-AIDL) for simultaneously learning an attribute-semantic and identity-discriminative feature representation space transferrable to any new (unseen) target domain for re-id tasks without the need for collecting new labelled training data from the target domain.
Frustratingly Easy Person Re-Identification: Generalizing Person Re-ID in Practice
TLDR
This paper proposes a simple baseline for domain generalizable~(DG) person re-identification, to learn a \reid{} model from a set of source domains that is suitable for application to unseen datasets out-of-the-box, without any model updating.
Cross-View Asymmetric Metric Learning for Unsupervised Person Re-Identification
TLDR
The model aims to learn an asymmetric metric, i.e., specific projection for each view, based on asymmetric clustering on cross-view person images, and finds a shared space where view-specific bias is alleviated and thus better matching performance can be achieved.
Distilled Person Re-Identification: Towards a More Scalable System
TLDR
A Multi-teacher Adaptive Similarity Distillation Framework is proposed, which requires only a few labelled identities of target domain to transfer knowledge from multiple teacher models to a user-specified lightweight student model without accessing source domain data.
Adaptation and Re-identification Network: An Unsupervised Deep Transfer Learning Approach to Person Re-identification
TLDR
The proposed adaptation and re-identification network (ARN) performs unsupervised domain adaptation, which leverages information across datasets and derives domain-invariant features for Re-ID purposes.
End-to-End Deep Learning for Person Search
TLDR
An end-to-end deep learning framework to jointly handle pedestrian detection and person re-identification is proposed and a random sampling softmax loss is proposed to effectively train the model under the supervision of sparse and unbalanced labels.
...
1
2
3
4
5
...