Computer Vision Based Method for Shadow Detection

Shadows in images and videos can cause the failure of computer vision techniques’ performance, such as segmentation, tracking, object detection, and recognition. Therefore, we propose a unique shadow detection for images and videos using artificial intelligence techniques. Our system is effective and can process real-time frames with different textures and the formality of shadows. This proposed method aims to improve the performance of the assistive devices that are provided for visually impaired people. We ran the proposed algorithm on different scenarios and different settings. An overall accuracy of 92% was achieved as a promising result.