An intelligent object detection and classification framework for assisting visually challenged persons using deep learning and improved crow search optimization
Abstract According to an estimation, one billion persons are experiencing disabilities, so assistive technologies are developed, enhancing independence and accessibility. Significant developments have been made in assisting disabled people. Object detection (OD) and classification systems are effect...
Saved in:
| Main Authors: | , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Nature Portfolio
2025-08-01
|
| Series: | Scientific Reports |
| Subjects: | |
| Online Access: | https://doi.org/10.1038/s41598-025-15793-0 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Abstract According to an estimation, one billion persons are experiencing disabilities, so assistive technologies are developed, enhancing independence and accessibility. Significant developments have been made in assisting disabled people. Object detection (OD) and classification systems are effective computer technologies for image processing and computer vision (CV). It is mainly used to identify and describe objects such as vehicles, individuals, and animals from digital videos and images, which will be useful for older or disabled persons. Deep learning (DL) models demonstrate to be more expert in resolving OD defects. However, DL techniques are extensively utilized to perceive, track, and identify in real-time objects met during navigation in an indoor environment. This study proposes a Hybrid DL Model for Object Detection and Classification Using an Improved Crow Search Algorithm (HDLMODC-ICSA) method. The HDLMODC-ICSA method primarily focuses on an accurate and real-time object recognition method to assist visually challenged persons. In the initial stage, the image pre-processing stage utilizes median filtering (MF) to remove noise or distortions and make the image more transparent. Furthermore, the OD process employs the Faster R-CNN model to generate precise region proposals and detect objects within images efficiently. Moreover, the HDLMODC-ICSA technique employs the Improved LeNet-5 model to extract meaningful and discriminative features from the identified regions. The hybrid of the attention-based stacked bi-directional long short-term memory (ABS-Bi-LSTM) technique is used for OD and classification. Finally, the hyperparameter selection of the ABS-BiLSTM model is performed by implementing the improved crow search algorithm (ICSA) model. The efficiency of the HDLMODC-ICSA approach is validated by comprehensive studies using the Indoor objects detection dataset. The comparison study of the HDLMODC-ICSA approach demonstrated a superior accuracy value of 99.59% over existing techniques. |
|---|---|
| ISSN: | 2045-2322 |