A Two-Stage YOLOv8 Approach for Waste Detection and Classification in Cognitive Cities

EasyChair Preprint 14850

Abstract

Waste, as a primary cause of visual pollution, not only impacts public health but also has significant economic implications, particularly in tourism. Visual pollution from waste or trash encompasses various types that require classification. Cognitive cities are beginning to develop automatic systems to classify these types, but the task is challenging due to the similarity among different types of waste and the common features of most elements. To address this issue, we propose an innovative approach using the YOLOv8 object detection model to detect 16 different types of trash. The proposed approach is compared to the traditional YOLOv8 to evaluate its performance. The results demonstrate the potential of the modified YOLOv8 approach, particularly when applied to larger image sizes achieving a notable improvement in F1-score, underscoring the viability of the proposed approach.

Keyphrases: Computer Vision and Pattern Recognition, Trash Detection and classification, Visual Pollution, cognitive cities, computer vision, deep learning, object detection

@booklet{EasyChair:14850,
  author    = {Ahmad Nayfeh and Sadam Al-Azani and Hussein Samma},
  title     = {A Two-Stage YOLOv8 Approach for Waste Detection and Classification in Cognitive Cities},
  howpublished = {EasyChair Preprint 14850},
  year      = {EasyChair, 2024}}