Positively identifying species using convolutional neural networks and hypernetworks to aid wildlife conservation efforts

Aditya Radhakrishnan, India 13-15

This project aims to solve the problem in wildlife conservation of positively identifying species by using CNNs to identify images of species. For endangered species, there is often an insufficient number of training examples, so the use of a novel method of utilizing hypernetworks was developed to solve the problem. These neural networks can help identify images of organisms and their remnants, such as feces and footprints. They can be deployed offline to be used in remote areas. This application can be useful and effective in curbing poaching and illegal trafficking, monitoring wildlife population, and diversity analysis.

Show more