Multi Scale Adaptive Super Resolution Person Re-Identification(MSA-SR Person Re-Identification)

Intelligent video surveillance systems,automated entry and retail systems at theme parks,passenger flow monitoring at airports,behavior analysis for automated driving are a few applications where detection and recognition of persons across a camera network can provide critical insights.Now a day,widespread networks of cameras are being used in various public places like railway stations,airports,hospitals,shopping malls,etc.which cover large areas and have non-overlapping viewpoints and thus provide a huge amount of relevant data.To utilize this data effectively for public safety applications,we cannot rely on manual monitoring and thus need efficient automated systems,which can track a person's activity across multiple cameras.Person reidentification(PREID)is an elementary task for this.It is an important computer vision task,which aims to identify a person from a set of gallery images captured under different cameras,or different timestamps under a single camera.This task assumes that the subject of the probe image belongs to the gallery;that is,the gallery contains this person.This scenario is also called closed-set identification.However,the task of Reid is quite challenging,problems like scales and resolution variation,variations in pose,changes in viewpoints,light intensity,blur background,and occlusion make it a non-trivial.Moreover,with limited facial information,soft-biometrics and biometrics-based technologies are invalid.In this thesis,our focus is on solving the scale and resolution variation problem and improving the overall performance of person re-Identification.In real-world surveillance systems,the person images captured by the camera network consists of various low-resolution(LR)images.It creates a resolution mismatching problem when compared against high-resolution images of a targeted person.It significantly affects the performance of person re-Identification.We call this problem as Low-Resolution Person re-identification(LR PREID).An efficient strategy would be to exploit image super-resolution(SR)with Person re-identification as a mutual learning approach.In this thesis,we propose a novel method MSA-SR-PREID to solve this problem.The model takes low-resolution images on different resolutions and resized them to pre-defined fixed resolution.The design of the super-resolution network consists of ESRGAN and the de-Noising module to generate super-resolution images.The SR images are later passed to the re-identification network to learn the unique descriptors to recognize a person identity.The performance of this model has been evaluated on four competitive benchmarks,MLR-VIPe R,MLR-Duke MTMC-re ID,VRMSMT17,and VRMarket1501.The comparison with similar state-of-the-art demonstrates the superiority of our model.