sdmay19-10 • Distributed Wild Bird Surveillance and Recognition System

This theme uses "root" links to dependencies and assets (CSS, JS, images, etc.). Place all theme files and directories at the root of your web server for this page to render as intended.

Senior Design Team sdmay19-10 • Distributed Wild Bird Surveillance and Recognition System

Project Plan
Design Document

Second Semester Report

Link to presentation Final Report

Problem Statement

Bird identification is a more complicated topic than one would think. As an example, can you identify the species of the four following birds?

In the clockwise order from the top left, the birds are a male cardinal, female cardinal, another female cardinal, and a juvenile cardinal. Four seemingly different birds, but they all happen to be the same species. Now imagine these differences happening for all the birds across North America. Because of this, even for a hobbyist, bird identification can be challenging. The task at hand to provide a solution to this problem is to create a self-contained system that can detect and classify bird species in realtime in our client’s backyard. The current consumer products available are limited to nature cameras that are constrained by basic video streaming and photo capture capabilities that are not able to classify birds. To do this, we will use an embedded GPU platform with a convolutional neural network for real time object detection and classification. Our model will be trained and tested with two already available North American bird datasets: NABirds V1 and Caltech-UCSD Birds 200. As a result of this project, we will have delivered a well-documented HW/SW product to our client that will successfully identify North American birds in real time.