Early and precise diagnosis of plant diseases is critical since the agriculture industry plays a key role in maintaining food security. For these kinds of applications, object identification and picture classification algorithms have greatly evolved with the introduction of deep learning. In order to identify plant diseases more effectively, this research investigates the use of pre-trained models based on Convolutional Neural Networks (CNNs), namely MobileNet, DenseNet121 and VGG16. We used a large dataset of agricultural images for our experiments. The models’ performance was assessed by looking at their classification accuracy and loss. According to our research, MobileNet outperformed VGG16 and DenseNet121, which obtained an accuracy of 97.71%. This study highlights the possibility for improved results when using Densenet121 with Adam optimization.