Dermatological infectious diseases pose a significant public health concern due to their highly contagious nature, often characterized by painful sores, fluid-filled blisters, flesh-colored bumps, and itching. Despite their distinct visual symptoms, diagnosing these diseases can be challenging due to their phenotypical similarities and overlapping clinical presentations with other skin conditions. In this paper we introduce a novel and diverse skin lesion dataset comprising patients from India, focusing on prevalent infectious skin conditions such as herpes zoster, herpes simplex, molluscum contagiosum, and non-viral skin disorders. These conditions are particularly common in this geographical region. Furthermore, we test Deep Learning models using different augmentation techniques, analyze the performance, and evaluate several metrics on different deep models using image augmentations. By assessing the performance of these models and analyzing several metrics across different augmentation methods, our findings demonstrate the capability of deep-learning models in classifying skin images and such computational techniques can be used to enhance healthcare accessibility and effectiveness, in resource-constrained settings like India.