A smart assistive system for visually challenged people through efficient object detection using deep learning with...

Area of Science:

• Computer Vision
• Artificial Intelligence
• Assistive Technology

Background:

• Visual impairment presents significant challenges in performing everyday tasks.
• Accurate object detection is crucial for developing effective assistive technologies for the visually impaired.
• Current methods require enhancement for reliable object recognition in diverse environments.

Purpose of the Study:

• To propose a Smart Assistive System for the Visually Challenged People through Object Detection Using the Tunicate Swarm Algorithm (SASVCP-ODTSA).
• To enhance object detection accuracy and classification performance for assisting visually impaired individuals.
• To automatically detect and classify objects in images to aid daily activities.

Main Methods:

• Image pre-processing using Median Filtering (MF) for noise reduction.
• Object detection utilizing the YOLOV8 method.
• Feature extraction with the CapsNet model.
• Object detection and classification using a Deep Belief Network (DBN) optimized by the Tunicate Swarm Algorithm (TSA).

Main Results:

• The SASVCP-ODTSA model achieved a superior accuracy of 99.58% on the Indoor Object Detection dataset.
• The Tunicate Swarm Algorithm effectively optimized the Deep Belief Network parameters for improved classification.
• The proposed system demonstrated robust object detection and classification capabilities.

Conclusions:

• The SASVCP-ODTSA system offers a significant advancement in assistive technology for the visually impaired.
• High accuracy object detection and classification are vital for enhancing independence and task completion for visually challenged individuals.
• The integration of MF, YOLOV8, CapsNet, and TSA-optimized DBN presents a powerful approach for object detection applications.