Executive Summary : | The goal of this project is to build a proof-of-concept tool for autonomous surveillance to detect and deter poaching of endangered animal species in forests of India. We propose a brain-inspired Deep Spiking Neural Network on the SpiNNaker, a neuromorphic hardware built with very-low-power ARM processors, that will be capable of visual learning, memorising and recall. The dynamic visual scene will be captured by a Dynamic Vision Sensor (DVS)-based camera, another low power neuromorphic device. After an intial period of training corresponding to a pre-determined set of environmental conditions, the network will adapt and act by recalling prior experience, alleviating the need for human intervention. We have been using the SpiNNaker neuromorphic computer to build spiking neural network models, computed in real time at a resolution of 1 millisecond. SpiNNaker server is available for academic research free of cost, either directly from the University of Manchester, or via the Human Brain Project (funded by European Union Horizon 2020) portal; further, boards with 48 SpiNNaker chips are available on loan as well as for sale at affordable prices. DVS-based camera will be available on loan from Teresa Serrano-Gotarredona's lab at the Instituto de Microelectronica de Seville (IMSE) Spain. We will be using the interface of DVS and SpiNNaker to build our low power device for forest surveillance and monitoring. The project outcome will in the long run reduce risk to human life, and help in maintaining the ecology of forests. |
Co-PI: | Dr. Sujith Thomas Birla Institute Of Technology & Science Pilani, Goa,Bits-Pilani K.K. Birla Goa Campus, Nh 17b Bypass Road, Zuarinagar, Sancoale,Goa,South Goa-403726 |