| Autor: |
Bhatlawande, Shripad, Katkalambekar, Varad, Singh, Khushi, Kulkarni, Chinmay, Madake, Jyoti, Shilaskar, Swati |
| Předmět: |
|
| Zdroj: |
AIP Conference Proceedings; 2023, Vol. 2754 Issue 1, p1-12, 12p |
| Abstrakt: |
The goal of this paper is to develop an assistive device for the visually impaired that will allow them to move more freely in unfamiliar indoor situations by recognizing household objects that may hinder their route. This solution focuses on safe navigation in unfamiliar interior environments. Human-assisted mobility and assistive aids are expensive, especially for people residing in developing countries like India. These are not affordable for the population which has a per capita GDP of 1500$. This solution aims to provide cues for movement in real-time which can be initiated by a simple obstacle detection. The basic feature of obstacle detection is to give a better experience and also be pocket-friendly. This system implements KNN, Random Forest, Decision Trees, and SVM classifiers. The proposed system analyzed that the Random Forest classifier gave the best performance accuracy for indoor obstacle detection. This system has implemented a simple guidance system that gives a tactile impulse to the user for moving in the right direction. The real-time assistive solution demonstrated 98.3% accuracy for object detection and reliable execution on a low-power device. [ABSTRACT FROM AUTHOR] |
| Databáze: |
Complementary Index |
| Externí odkaz: |
|
