Eff2Net: An efficient channel attention-based convolutional neural network for skin disease classification

The primary layer of protection for vital organs in the human body is the skin. It functions as a barrier to protect our internal organs from different sources. However, infections caused by fungus, viruses, or even dust can damage the skin. A tiny lesion on the skin can grow into something that can cause serious health problems. A good diagnosis can help the person suffering from a skin disease to recover quickly. This research aims to develop a system for detecting skin diseases using a Convolution Neural Network (CNN). The proposed model named Eff2Net is built on EfficientNetV2 in conjunction with the Efficient Channel Attention (ECA) block. This research attempts to replace the standard Squeeze and Excitation (SE) block in the EfficientNetV2 architecture with the ECA block. By doing so, it was observed that there was a significant drop in the total number of trainable pa-rameters. The proposed CNN learnt around 16 M parameters to classify the disease, which is comparatively less than the existing deep learning approaches reported in the literature. This skin disease classification was per -formed on four classes: acne, actinic keratosis (AK), melanoma, and psoriasis. The model achieved an overall testing accuracy of 84.70%.

Automated summary

The skin serves as the primary protection layer for vital organs in the human body. This study proposes a system using a Convolution Neural Network (CNN) to detect and classify common skin diseases. By replacing certain modules and reducing trainable parameters, the proposed CNN achieved an overall testing accuracy of 84.70% for acne, actinic keratosis (AK), melanoma, and psoriasis.