Person Re-Identification Research via Deep Learning

Jian, 卢健 Lu; Xu, 陈旭 Chen; Maoxin, 罗毛欣 Luo; Hangying, 王航英 Wang

doi:10.3788/lop57.160003

Cited by 1 publication

(1 citation statement)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In Person Re-id, the similarities of the images of different pedestrians might be high due to illumination, change of background and change of body gesture. [8] conclude a global loss term based on higher order statistics that investigates the overall structure of the embedding space in order to optimize the local loss. The [8] summarized how to employ an integrated architecture that combined both deep and shallow neural networks to spontaneously collect feature and similarity metrics and to tune the neural networks via extensive triplet sampling.…”

Section: Distance Metric Learning (Ml)mentioning

confidence: 99%

Evaluation of deep learning-based algorithms on person re-identification

2023

ACE

View full text Add to dashboard Cite

At present, CCTVs and multifunctional cameras have been well equipped everywhere in our daily lives. However, most of the public sectors especially the department of public safety are still using the traditional approach, analysing these data by humans. This research presents a summary of the current technologies and algorithms based on deep learning in person Re-identification (Re-id), which aims to recognize a particular individual in several camera views. This paper firstly described the current challenges in using deep learning for person Re-id. Then, a brief introduction of standard person Re-id including feature learning and metric learning is given. Furthermore, the open world Re-id problem is examined from three aspects: person re-id with occlusions, unsupervised learning-based method, and cross-modal person re-Id. The author then evaluated the regularly used datasets and analysed the performance of the algorithms from recent times. Finally, some potential future directions of research for Re-id are proposed.

show abstract

Section: Distance Metric Learning (Ml)mentioning

confidence: 99%

Evaluation of deep learning-based algorithms on person re-identification

2023

ACE

View full text Add to dashboard Cite

show abstract

Person Re-Identification Research via Deep Learning

Cited by 1 publication

References 0 publications

Evaluation of deep learning-based algorithms on person re-identification

Evaluation of deep learning-based algorithms on person re-identification

Contact Info

Product

Resources

About