Coconut plantations are crucial role in the agricultural economy in tropical regions. The detection of Coconut Leaf Disease (CLD) is significantly challenged by the presence of pests and other diseases in farming. Therefore, early detection of diseases in coconut leaves is essential for maintaining crop health and maximizing yield. Hence, this study presents a comprehensive approach to the classification of coconut leaf images through a robust image processing integrated with transformer based Deep Learning (DL) technique. The coconut diseases and pest infestations dataset is applied pre-processing stage using a Guided Image Filter (GIF), which enhances image quality by preserving edges while suppressing noise. Following this, Active Contour Method (ACM) is employed for precise segmentation, enabling accurate delineation of leaf boundaries. To extract meaningful texture features, the Local Ternary Pattern (LTP) technique is applied, offering improved noise resistance and discriminative pattern compared to traditional texture descriptors. Finally, a lightweight yet powerful transformer-based deep learning (DL) model, the Mobile Vision Transformer (MobileViT), is employed to classify healthy and infected coconut leaves. Using Python software, the proposed framework achieved higher accuracy, F1-score, precision and recall of 96%, is accomplished when compared to other techniques.