Abstract
Person re-identification systems (person Re-ID) have recently gained more attention between computer vision researchers. They are playing a key role in intelligent visual surveillance systems and have widespread applications like applications for public security. The person Re-ID systems can identify if a person has been seen by a non-overlapping camera over large camera network in an unconstrained environment. It is a challenging issue since a person appears differently under different camera views and faces many challenges such as pose variation, occlusion and illumination changes. Many methods had been introduced for generating handcrafted features aimed to handle the person Re-ID problem. In recent years, many studies have started to apply deep learning methods to enhance the person Re-ID performance due the deep learning yielded significant results in computer vision issues. Therefore, this paper is a survey of the recent studies that proposed to improve the person Re-ID systems using deep learning. The public datasets that are used for evaluating these systems are discussed. Finally, the paper addresses future directions and current issues that must be considered toward improving the person Re-ID systems.
Highlights
Person Re-Identification (Person Re-ID) has recently attracted academic attention in the field of computer vision
Person Re-identification is a significant task in intelligent video surveillance systems
The performance was measured by cumulative match characteristic (CMC) metric and the mean average precision
Summary
Person Re-Identification (Person Re-ID) has recently attracted academic attention in the field of computer vision. The authors in [52] proposed the first model for handling RGB-Infrared cross-modality person Re-ID problem and introduced a new multi-modal Re-ID dataset called SYSU-MM01 It used three common neural network structures including one-stream, two-stream and asymmetric fully connected layer. The performance was measured by cumulative match characteristic (CMC) metric and the mean average precision (mAP) By comparing this model with other existing models, In RegDB dataset, the results at rank-1 to rank-20 ranged from 33.47 % to 67.52% (31.83% mAP). The first video-based person Re-ID model using deep learning was proposed by authors in [54] They introduced the recurrent convolutional network that used both color and optical flow features, the color described the appearance of the person, and optical flow described short-term motion, containing the person gait and other motion features. DukeMTMC-VideoReID [82]: it consists of 369,656 images for 702 identities for training and 445,764 images for 702 identities for testing, there are 408 identities as the distractors recorded from 8 cameras
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
