Deep Learning-based Convolutional Neural Network Model for Hair Diseases Detection

Bleaching, dying, straightening, curling, and other chemical treatments for hair are becoming increasingly common around the world as people's interest in hairstyles and hair colouring is increasing. As a result, hair has sustained significant damage that can be observed with the naked eye and by touching the hair texture. The chemicals applied to the hair can produce severe health issues such as skin cancer, migraine, and hair fall. Despite the dangerous consequences of hair colouring and treatments, people are still applying these chemicals. The hair disease detected at its early stages can lead to reducing hair loss and avoiding skin cancer and migraine. With the advancements in technologies, methods for early hair disease detection are also developing. In the proposed work, a hair disease dataset has been collected from Kaggle which is further implemented by using a convolutional neural network model. The results have been calculated at different epochs using two different optimizers namely, SGD and Adam which identified that the proposed model outperforms at epoch 85 with ADAM optimizer by achieving an accuracy rate of 95%. The model implemented with SGD has achieved the highest accuracy of 89% at epoch 50. This model has achieved better outcomes when compared with the other existing models.