Accurate plant species identification is vital for biodiversity conservation, sustainable agriculture, and scientific research. Traditional manual identification methods are time-consuming, labor-intensive, and prone to errors, highlighting the need for automated and efficient classification systems. This paper presents a deep learning-based ensemble technique that uses the VGG16, VGG19, and ResNet50 models to improve the accuracy of plant species classification system. The proposed approach makes use of data pre-processing, augmentation, feature extraction, and adaptive ensemble learning techniques. The ensemble technique combines model predictions using weighted average and max voting to optimize performance on the Leafsnap datasets. The ensemble technique achieves an accuracy of 98.53%. Experimental results indicate that the ensemble method surpasses individual deep learning models, offering an efficient and highly accurate solution for automated plant species identification.