Irfan Yaqoob

Low-cost air quality sensors are more accessible, their data can be unreliable. Without proper calibration, the readings can mislead decision-makers, especially when it comes to environmental regulations. Our research focuses on using machine learning techniques to calibrate data from PurpleAir sensors to align with accurate Federal Monitors. The motivation is to improve their accuracy and make widespread air quality monitoring more feasible.

We argue that accurate person re-identification is a vital problem for urban public monitoring systems in the smart city context. Since images captured from different cameras have arbitrary resolutions resulting in resolution mismatch, this work proposes a model that takes arbitrary images and converts them to a pre-defined fixed resolution. The model then passes the images to a super-resolution network, producing high-resolution images. We employ a feedback network to generate more realistic super-resolution images, which are fed to the re-identification network to acquire a unique descriptor to disclose the person’s identity. We outperformed in all measures against other state-of-the-art methods.

Person re-identification is a vital part of the computer vision and plays an important role, especially for surveillance applications, e.g., intelligent video analysis, forensic, behavior analysis, and robotics. Mostly it is assumed that gallery and target images have the same resolution. However, this assumption fails in real-world re-identification because images taken by security cameras are mostly low-resolution and have pose variations, background clutters, and occlusions. So it makes the re-identification a challenging task. We proposed an efficient deep learning approach to address the low-resolution problem that regenerates low resolution images to uniform high-resolution images. Afterward, these high-resolution images are passed to the re-identification network to get the final output. We employ a feedback network to convert low-resolution images to high-resolution. Our network's performance is tested on two datasets, MLR-VIPeR, MLR- DukeMTMC-ReID, and our model achieved superior results compared to the other algorithms.