<p>Paddy production is facing continues constraints significantly due to wide range of leaf diseases that affect yielding and also grain quality. It leads the economic losses to the farmers and it is threat to the food security across the paddy cultivating countries. In this regards, it is very difficult to ensure reliable identification of diseases at an early stage, where visual inspection remains the primary method used by farmers. Recent deep learning (DL) models have shown promise in automating this task, but many struggle to capture both the fine lesion patterns and the broader structural cues present in real paddy leaf images. To address this gap, this proposed research work illustrates a hybrid and efficient deep learning transformative agricultural framework that integrates the Swin Transformer (ST) with the Convolutional Vision Transformer (CVT) approaches. This framework combines hierarchical attention features from Swin Transformer along with convolution aware token representations from Convolutional Vision Transformer, producing a fused feature space that supports robust classification effectively. In this work, the Paddy Doctor dataset, a publicly available benchmark dataset hosted on the Kaggle repository is used for performance evaluation. This proposed hybrid ST-CVT model achieves 98.6%, performing better than standard CNN and other transformer models. These results indicate that combining global and local feature reasoning can substantially improve automated paddy disease diagnosis which will helpful for early awareness from formers to reach aimed productivity in agricultural cultivation.</p>