Potato cultivation faces significant threats from leaf diseases like early and late blight, impacting crop yields and food security, necessitating accurate early detection. Traditional visual inspection methods are subjective and time-consuming, while laboratory techniques are often inaccessible to small farmers. This paper presents an improved approach for potato leaf disease classification combining deep learning with nature-inspired optimization. The proposed method utilizes EfficientNet-B0 for feature extraction and Binary Chimp Optimization (BCO) for feature selection, followed by K-Nearest Neighbors (KNN) classification. The methodology addresses these challenges by automating disease detection through computer vision and machine learning. Experimental results demonstrate superior performance with 98.6% accuracy on the Plant Village dataset, outperforming existing methods. The BCO algorithm effectively reduces feature dimensionality by 20% while improving classification accuracy. Comparative analysis with state-of-the-art techniques confirms the robustness of the approach across different validation set configurations. The integration of deep feature extraction with optimized feature selection offers a balanced approach between model complexity and performance.