Wild animal species detection using deep convolution neural network
Arshita Verma, Vishu Sangwan, Neha Shukla · 2021
Monitoring animals in forest one to one is very tedious job so technology have evolved which include installing cameras in animal living areas and acquire videos and pictures. But the pictures acquired from installed camera are always not good most of the time pictures are blurred and without animals, so it’s hard to detect animals, resulting in doubt and error. For solving this obstacle, we proposed a database of installed camera network captured images with multilevel graph of different animal species stored in database in spatiotemporal domain and pictures with animal are captured from forest are compared with the database, identifying whether a graph matches the animal or anything else. We created a model for detecting animals based on oneself training by the method explained in the paper. The technology is used for classification using machine learning (ML) and Artificial intelligent (AI) algorithms, hold up vector machine, k-mean closest neighbor, together with group tree. From the experimental results it is shown that, the proposed system accurately classifies wild animals up to 91% accuracy.