College - Author 1

College of Engineering

Department - Author 1

Computer Science Department

College - Author 2

College of Engineering

Department - Author 2

Computer Science Department

College - Author 3

College of Engineering

Department - Author 3

Computer Science Department


Franz Kurfess, College of Engineering, Computer Science Dept. ;Sumona Mukhopadhyay, College of Engineering Computer Science Dept.

Funding Source

The Noyce School of Applied Computing




The goal of this project is to be able to accurately detect and count livestock in footage captured by a drone in real time. The main problems with this arise from the fact that a drone can only carry limited computing resources, and hashing is conventionally thought of as a great method of doing image classification very quickly and thus even on low-power devices. In this project, we use both a Faster-RCNN, which is a state-of-the art object detection model as a benchmark to develop a hashing model that can perform a similar task much more quickly. These two models provide a trade-off between accuracy and speed, where the Faster-RCNN is more accurate and gives precise locations of the livestock in the image, while the hashing is significantly faster but is less accurate and only provides the number of livestock in the image. Given that the dataset is very limited in quantity, we also build a generative network to create more images for the model to train on so that it has a more diverse set of hash codes to reference.




To view the content in your browser, please download Adobe Reader or, alternately,
you may Download the file to your hard drive.

NOTE: The latest versions of Adobe Reader do not support viewing PDF files within Firefox on Mac OS and if you are using a modern (Intel) Mac, there is no official plugin for viewing PDF files within the browser window.