In India, agriculture serves as the primary income source for the majority of the population. Identifying crop diseases is a critical factor in mitigating production losses. To address this, deep learning techniques, specifically utilizing pre-trained Convolutional Neural Network (CNN) models such as ResNet-50, VGG-16, MobileNetV2, and InceptionV3, are employed for the detection of plant diseases. This study involves different key stages including dataset creation, preprocessing, data augmentation, and classification. The dataset comprises 3725 images of cotton plant leaves distributed across 11 classes. Here, the model performance is assessed based on classification accuracy, with ResNet-50 achieving the highest accuracy at 99.8% among the four approaches.