Accurate diagnosis of plant diseases is vital for crop survival and growth directly influencing the agricultural economy. However, the limited availability of plant pathology specialists presents a significant challenge. This research proposes an automated system utilizing advanced deep learning techniques to identify plant diseases from images. Traditional deep learning models often require large datasets, which are typically unavailable due to their limited coverage and different sample sizes for various diseases. To address this a Meta Learning approach specifically Few Shot Learning has been applied which excels in scenarios with limited data by rapidly adapting to new classes with minimal samples. Furthermore, a Multi-scale architecture has also been implemented to enhance feature extraction and improve performance on small datasets. The proposed approach demonstrates substantial improvements in accurately diagnosing plant diseases with limited data, offering a practical solution for farmers to reduce economic losses caused by undiagnosed or misdiagnosed plant conditions.