Accurately classifying tree species is a complex task influenced by multiple factors, from overall tree silhouette to fine-grained diagnostic features. In this study, we propose a hierarchical classification framework that enhances species identification accuracy through a multi-stage pipeline. The pipeline consists of three key classifiers: an Image Type Classifier (TC) to categorize input images, a Tree Silhouette Classifier (SC) to analyze overall form, and a Diagnostic Features Classifier (DC) to assess detailed attributes. The outputs of these classifiers are integrated using aggregation techniques to refine predictions. To evaluate the effectiveness of this framework, we benchmark state-of-the-art deep learning models including ResNet, EfficientNet, MobileNetV2, and DenseNet on our unique dataset of annotated Polish trees.