Snake Species Identification using Digital Image Processing

This research proposes an application of classifying the snake species using an image processing approach. We carried out the classification with the aid of Inception-V3, a trained Convolutional Neural Network (CNN) model. It retrains and trains the images of two snake species: Malayan Pit Viper (from venomous species) and Reticulated Python (from non-venomous snake species). Snakes are reptiles that contribute largely to nature as they function as a predator to control the population of jungle and field rats. Even though most snake species in Malaysia are non-venomous and give no harm to humans, some species can cause harm and danger to humans. Consequently, each year, local Malaysians or visitors to Malaysia are exposed to snakebites. In order to give the right treatment to the victim, it is important to first know the snake species in order to get the right anti-venom. Wrong identification of snake species may lead to the misgiving of antivenom. The present research manages to achieve 90% of the accuracy of the 40 images of snakes in helping doctors or authorized personnel, such as the Jabatan Bomba dan Penyelamat Malaysia (JBPM) and Angkatan Pertahanan Awam Malaysia (APM), to recognize the snake species before giving the victim the right treatment as well as catching the snake.