Abstract: Highlights•Introduces a compact EfficientNet model with twice the speed of the original version.•Maintains similar accuracy on ImageNet while improving efficiency for mobile deployment.•Expands HAM10000 dataset by adding a new class, creating the HAM10000-Plus dataset.•HAM10000-Plus enables better model performance in real-world skin disease diagnosis.•Evaluates models on ImageNet, HAM10000, and HAM10000-Plus for skin disease diagnosis.•Compares proposed model with other state-of-the-art models for accuracy and efficiency.
Loading