Autism Spectrum Disorder is categorized by a variety of traits, such as difficulties interacting socially, diverse learning styles, a tendency towards routines, difficulties communicating normally, and unique ways of processing sensory data. These kids' growth may be significantly aided by early intervention and the right supports. The diagnosis and screening of ASD have, however, run into significant challenges. According to the literature, specific retinal characteristics are strongly linked to ASD. In this work, we looked into applying deep learning techniques to retinal pictures in order to improve classification accuracy. For the classification job, the pretrained Convolution Neural Network (CNN) model was employed. 1,920 retinal images made up the dataset that remained utilised to test the model, which stood obtained through the Kaggle stage. Accuracy was the common evaluation metrics cast-off to assess the effectiveness of the deep learning model and achieved an accuracy result of 84.87%