Inadequate agricultural picture classification complicates crop monitoring, increasing the necessity for new technologies to improve crop identification and classification. Accurate and timely identification of cotton plant health is vital for effective disease management and pest control. This research goals to design a tool to classify cotton plant images into different categories, such as healthy, diseased, and pest-infested the purpose of this work is to make a tool for knowing and categorizing cotton plant photos based on features such as leaf shape, colour, and texture. The system utilizes a deep learning model constructed on Faster R-CNN with ResNet-50 and ResNet-101 architectures, which it has proven model of effective in a large amount of image recognition tasks. However, due to the difficulty and diversity of plant images, traditional approaches often lack the accuracy and efficiency required. To overcome these difficulties, this study hires deep learning algorithms to detect the diseases, stages and classify cotton plant images. The research began with the collection of samples nearly 250 cotton plant photos, which were separated into the ratio of 80:20 to train and test the datasets. Training data comprised 100, 150, and 200 images respectively, for each evaluation of trained data, test data for each experiment included 50 and 80 images, with 30 images from the training dataset included in the 50-images on the test set. Each training process involved up to 2000, 3000, and 4000 steps, each with a corresponding total loss stricture. Based on the evaluation results, the Faster R-CNN ResNet-50 and ResNet-101 Architecture that achieved average F1 Scores of 65% and 79%, respectively. The Faster R-CNN ResNet-101architecture model that demonstrated superior performance. The best F1 Score has Faster R- CNN ResNet-101.The best performance was achieved using the Faster R-CNN ResNet-101 model, which demonstrated higher accuracy in classifying cotton plant images.