Paper Key : IRJ************890
Author: Manikandan
Date Published: 03 Mar 2025
Abstract
Abstract Autism Spectrum Disorder (ASD) is a complex neurodevelopmental condition marked by difficulties in social interaction, communication, and behavior. Early and accurate identification of ASD is essential, as it enables timely interventions that significantly improve developmental outcomes for individuals affected. Conventional diagnostic methods often depend on subjective assessments, which result in variability and delays in diagnosis. To address these challenges, this project proposes detecting autism through a Cookoo Search-optimized Deep Convolutional Neural Network (DCNN) classification. The process starts with acquiring input images, followed by preprocessing to enhance image quality through histogram equalization and reduce noise using a Gaussian filter. Data augmentation techniques are utilized to expand the dataset's diversity, boosting the models robustness and generalization. The Scale-Invariant Feature Transform (SIFT) is applied for feature extraction, allowing the system to capture important visual patterns from the images. For image retrieval and classification, a Deep Convolutional Neural Network (DCNN) is employed, optimized using the Cookoo Search Algorithm, which refines the classification process for improved accuracy. This combined approach enhances detection precision by improving feature extraction and optimizing network weights. Lastly, the predicted outcomes are deployed through an interface built on Streamlit , enabling real-time classification and interpretation. The proposed model shows considerable improvements in accuracy and efficiency, providing a promising solution for ASD detection. The purpose of this study is to detect autism from facial images using a deep learning model. To accurately identify autism in children, the pre-trained PSO and Cookoo Search Optimized Dcnn model, SIFT are used as feature extractors. The suggested models were trained using a publicly available dataset from Kaggle that included 3014 images of children characterized as autistic and non-autistic. The models yielded accuracies of non autistic image of accuracy, precision, recall of 96.%, 94 %, and 98%, respectively. This project is implemented using Python.Keywords: ASD - Autism Spectrum Disorder, SIFT - Scale-Invariant Feature Transform, DCNN - Deep Convolutional Neural Network, WHO - World Health Organization, CSA - Cookoo Search Algorithm ,PYPI - Python Package Index.
DOI Requested