Stair Detection and Classification Using Deep Neural Network for the Visually Impaired

Main Article Content

Ashwini Gaikwad, Dr. Vinaya V. Gohokar, Dr. Rupali Kute, Bhakti Paranjape

Abstract

Stair case detection is very challenging task for visually impaired people. It plays very important role to avoid accidents. Detecting stairs are comparatively easy but classification of stair case as up and down is a complex task. The literature survey more focuses on classifying up stair cases. However, detection of downstairs is also more important for visually impaired. The paper presents work done in the field of stair case detection for visually impaired people custom dataset of stairs up and down is prepared. The images are taken under various light conditions and background. It mainly focuses to detect up and down stairs using a different pertained model and comparative performance analysis is presented. The fine-tuned VGG-19 pretrained deep learning model gives better performance.

Article Details

Section
Articles