The individuals across the world with autism spectrum disorder (ASD) can vary widely in their symptoms and levels of impairment. ASD typically appears in early childhood and lasts throughout a person’s life. The combination of genetic and environmental factors causes challenges in social interaction, communication, restricted, and repetitive behaviors. Early detection and intervention are crucial for individuals with ASD, so this project introduces an innovative methodology for detection of autism using Kaggle’s facial image dataset. The Haar cascade technology along with the incorporation of multiple algorithms in deep learning enhances the model’s efficacy, enabling it to discern complex patterns indicative of ASD across varied age groups. The deployment on edge devices underscores the practical applicability of the proposed solution, contributing to the advancement of accessible and efficient healthcare technologies. This method outperforms than traditional methods such as VGG16, ResNet 50, Inception V3, and Convolution Neural Network (CNN) in terms of accuracy and efficiency.