VPP: Visual Pollution Prediction Framework Based on a Deep Active Learning Approach Using Public Road Images

Visual pollution (VP) is the deterioration or disruption of natural and man-made landscapes that ruins aesthetic appeal an area. It also refers to physical elements limit movability people on public roads, such as excavation barriers, potholes, dilapidated sidewalks. In this paper, end-to-end visual prediction (VPP) framework based a deep active learning (DAL) approach proposed simultaneously detect classify pollutants from whole road images. The architected around following steps: real VP dataset collection, pre-processing, DAL for automatic data annotation, splitting well augmentation, simultaneous detection classification. This designed predict localization it into three categories: A with 34,460 images was collected various regions across Kingdom Saudi Arabia (KSA) via Ministry Municipal Rural Affairs Housing (MOMRAH), used develop fine-tune artificial intelligence (AI) use five AI predictors: MobileNetSSDv2, EfficientDet, Faster RCNN, Detectron2, YOLO. VPP-based YOLO outperforms competitor predictors superior performance at 89% precision, 88% recall, F1-score, 93% mAP. plays crucial role in automatically annotating supporting VPP improve by 18% 27% 25% able distinct annotated strategy. technique applicable real-time monitoring applications.