A Contribution to the Prevailing Visibility Prediction Based on Neural Network Approach

This paper explores visibility prediction from images through the integration of object detection and machine learning. The proposed approach utilizes the You Only Look Once (YOLO) neural network to detect and classify objects within images, which are subsequently used for visibility prediction using a Multilayer Perceptron (MLP). The aim is to propose an innovative approach for visibility prediction based on visual data and highlight YOLO’s performance in static scenarios. The work compares the neural network approach with a simple method that only identifies the furthest objects. The paper demonstrates the effectiveness of YOLOv7 in this specific use case. Furthermore, it discusses potential improvements and approaches to enhance visibility prediction accuracy.