This paper presents a deep learning-based approach to detect mental stress from electrocardiogram (ECG) signals. The proposed method employs data augmentation and a shallow deep learning architecture combined with convolutional neural networks (CNNs) and long short-term memory (LSTM) networks. The model was trained and validated using 132 records collected from 22 healthy subjects. The proposed approach achieves an accuracy of 75%, sensitivity of 70.37%, specificity of 84.62%, precision of 90.48%, and f1-score of 79.17% in detecting mental stress from ECG signals. This study highlights the significance of using a combination of CNN and LSTM networks to achieve ECG-based stress classification. The proposed method has potential applications in the field of mental stress monitoring and management.