Venomify: Automated Classification of Venomous and Non-Venomous Snake Species Using Deep Learning

The identification of venomous and non-venomous snake classification play a vital role to enhance public safety and wildlife management. Traditional methods often rely on expert knowledge and physical communication, which can be time-consuming and prone to error. Therefore, an advocate solution is required to automate the identification process. In the present paper, a novel approach “Venomify”, a deep learning framework is used to identify venomous and non-venomous snake species. The proposed methodology involves the dataset preparation and preprocessing that includes the collection of images using Flickr API from Google. An image augmentation algorithm is also proposed to maintain the balance of both the classes of venomous and non-venomous snakes that helps to enhance the recognition accuracy of Venomify framework. By employing modified DenseNet169, a deep learning model is proposed that is capable of distinguishing venomous and non-venomous snake species from various snake images. The experimental results illustrates the supremacy of the proposed work and achieved 97.35% accuracy, precision of 97.35%, recall of 97.37% and F1-score of 97.35%. Venomify can serve as a powerful tool for the identification of the category of snake i.e. either venomous or non-venomous, offering rapid and reliable assessments that can aid in education, wildlife management, and emergency response.