One of the most prevalent types of cancer globally, skin cancer underscores the urgent need for advanced, scalable, and precise detection devices. Utilizing the International Skin Imaging Collaboration dataset, a large repository of annotated dermoscopic images, this work introduces a new approach to skin lesion diagnosis and classification using state-of-the-art deep learning methods. Applying EfficientNet, an extremely optimized neural network model, to feature extraction and categorization, the presented method avails itself of its capability of finding a trade-off between accuracy and computational frugality. Adaptive augmentation is performed through techniques such as geometric transformations and intensity scaling to maximize the diversity of training data. Graph-based segmentation is utilized for defining lesion boundaries, making feature extraction more accurate. For feature improvement, wavelet-based texture analysis is integrated with deep features to enhance diagnostic performance. The suggested model, EffiWave-Net, integrates EfficientNet for effective feature extraction with wavelet-based texture analysis and achieves superior performance in skin lesion diagnosis and classification. With 94.1% accuracy, preciseness, which is 0.96, recall, which is 0.94, and F1 score, which is 0.95, the model surpasses existing methods and attains remarkable performance. The union of EfficientNet with state-of-the-art feature representation methods maximizes the strength of this solution in overcoming the limitations of inter-class variability and data imbalance. This work demonstrates the power of integration through the coupling of novel architectures and creative preprocessing methods in dermatology diagnostics.