This study aims to diagnose autism in children using deep learning techniques by analyzing facial features. The authors employed three pre-trained Convolutional Neural Network (CNN) models—VGG16, VGG19, and EfficientNetB0—as feature extractors and binary classifiers. The models were trained on a publicly available dataset from Kaggle, which included 3014 images of children, 1506 of which were autistic and 1506 were non-autistic. The models achieved accuracies of 84.66%, 80.05%, and 87.9%, respectively. The study highlights the potential of deep learning in early diagnosis of autism, particularly through facial feature analysis, and suggests that transfer learning can enhance the efficiency and accuracy of these models. The research also discusses the importance of hyperparameter tuning and data augmentation to improve model performance. The findings indicate that the EfficientNetB0 model, when trained with Adamax as the optimizer and a learning rate of 0.001, achieved the highest accuracy of 88.33% and an AUC of 95.44%.This study aims to diagnose autism in children using deep learning techniques by analyzing facial features. The authors employed three pre-trained Convolutional Neural Network (CNN) models—VGG16, VGG19, and EfficientNetB0—as feature extractors and binary classifiers. The models were trained on a publicly available dataset from Kaggle, which included 3014 images of children, 1506 of which were autistic and 1506 were non-autistic. The models achieved accuracies of 84.66%, 80.05%, and 87.9%, respectively. The study highlights the potential of deep learning in early diagnosis of autism, particularly through facial feature analysis, and suggests that transfer learning can enhance the efficiency and accuracy of these models. The research also discusses the importance of hyperparameter tuning and data augmentation to improve model performance. The findings indicate that the EfficientNetB0 model, when trained with Adamax as the optimizer and a learning rate of 0.001, achieved the highest accuracy of 88.33% and an AUC of 95.44%.