TEEGALA KRISHNA REDDY ENGINEERING COLLEGE

Department of Computer Science & Engineering

HEAD OF THE DEPARTMENT PROJECT

CO-ORDINATER
Dr. CH. V. PHANI KRISHNA Mrs. P. SWETHA (Asst prof)
Professor Mrs. N.V.N SOWJANYA
(Asst prof)
Identification of Autism in Children Using Static
Facial Features and Deep Neural Networks

INTERNAL GUIDE NAME OF THE STUDENTS

Mr. K. Devadas A. Akshay(20R91A0529)
Assistant Professor D. Tarun(20R91A0544)
A. Shanmukha
Chandra(20R91A0551)
 ABSTRACT
Autism spectrum disorder (ASD) is a complicated neurological developmental
disorder that manifests itself in a variety of ways. The child diagnosed with ASD and their
parents’ daily lives can be dramatically improved with early diagnosis and appropriate
medical intervention. The applicability of static features extracted from autistic children’s
face photographs as a biomarker to distinguish them from typically developing children is
investigated in this study paper. We used five pre-trained CNN models: MobileNet, Xception,
EfficientNetB0, EfficientNetB1, and EfficientNetB2 as feature extractors and a DNN model
as a binary classifier to identify autism in children accurately. We used a publicly available
dataset to train the suggested models, which consisted of face pictures of children diagnosed
with autism and controls classed as autistic and non-autistic. The Xception model
outperformed the others, with an AUC of 96.63%, a sensitivity of 88.46%, and an NPV of
88%. EfficientNetB0 produced a consistent prediction score of 59% for autistic and non-
autistic groups with a 95% confidence level.
 EXISTING SYSTEM
1. Traditional diagnostic methods: The current approach to diagnosing autism
relies on clinical evaluations, behavioral assessments, and interviews.
2. Subjective judgment: Diagnosis is often based on subjective judgment,
leading to potential delays and variability in identification.
3. Limited access to specialized tools: Shortages of skilled professionals and
access to specialized diagnostic tools can hinder timely diagnoses.
4. Lack of quantitative measures: The existing system may lack precise and
quantitative measures, relying on observational and behavioral criteria.
DRAWBACKS
 Delay in Diagnoses
 Limited Quantitative Measures
 Dependence on Observational Criteria
 Resource Intensive
 Limited Integration of Technological Advances
 Potential for Bias
PROPOSED SYSTEM
 Deep Learning Approach: Introduces a novel system using deep neural
networks, specifically pre-trained CNN models (MobileNet, Xception,
EfficientNetB0, EfficientNetB1, EfficientNetB2).
 Facial Feature Biomarkers: Utilizes static features extracted from facial
photographs as biomarkers for autism identification.
 Improved Accuracy: Aims to enhance accuracy and efficiency in autism
diagnosis compared to traditional methods.
 Xception Model Performance: Demonstrates superior performance of the
Xception model with an AUC of 96.63%, sensitivity of 88.46%, and NPV of
88%.
 Machine Learning Integration: Introduces a machine learning-based
approach for more objective and quantitative measures in autism diagnosis.
 Potential for Early Detection: Provides a potential alternative for early
detection, emphasizing the significance of facial features as reliable indicators.
Thankyou