The majority of nations depend heavily on agricultural exports to finance their economies. Damage to agricultural output from plant disease (PD) can result in catastrophic losses. Early identification of PD is essential for minimizing losses in agricultural production and quantity. Signs of a PD are reflected in the plant’s leaves. Variability in crop species, crop disease signs, and environmental conditions make early diagnosis of illnesses affecting potato leaves challenging. To tackle this difficulty, machine learning (ML) approaches have been opted. Many deep learning (DL) methods have been established in recent years for identifying potato leaf diseases (PLD). In this research, we offer an approach for classifying diseases affecting potato leaves by employing DL and image segmentation techniques. Collecting data, preprocessing that data, segmenting an image, extracting features, and classifying that picture are the five main components of the suggested technique. The primary objective of this research is to use image processing methods to identify diseased potato plants utilizing the Plant Village image dataset (PVD). Diseased areas in images of potato leaves are separated using the K-means clustering technique. In order to categorize PLD, a deep neural network (DNN) model is used with Adam and categorical cross-entropy hyperparameters. With the suggested approach, classification accuracy attained is 98.52% for PLD detection. Our model successfully detected and classified leaf diseases in potato plants, as evidenced by experimental findings.