SBIR Phase I: Automated processing of images and video for wildlife conservation

Wildlife Imaging Systems Llc, Hinesburg VT

Investigators

Abstract

The broader impact of this Small Business Innovation Research (SBIR) Phase I project is to revolutionize wildlife field work and research. Many wildlife professionals already understand the potential for using video as a sensor, but without commercial software to turn video into data, its vast benefits remain untapped. It is important to monitor environmental effects and the impact of urban growth; billions of dollars are spent annually understanding how construction and infrastructure projects are impacting species and habitats. The state of practice is to send technicians into the field for observation. In many situations using a camera system to monitoring for wildlife is not only safer and more effective, but also much less expensive that using human labor. This project will reduce the cost, decrease the timeline, and yield better outcomes for these important conservation efforts, allowing the existing biodiversity to be preserved. This Small Business Innovation Research (SBIR) Phase I project will develop a web-based software-as-a-service to detect, track, and classify wildlife from video and images captured in the field. There is substantial potential for video to provide better data to help wildlife biologists perform their work, but the main barrier is the complexity. Using video systems to study wildlife requires advanced multidisciplinary knowledge; appropriate camera technology, recording methods, advanced video processing techniques and machine learning algorithms. There are currently no commercial-off-the-shelf hardware or software packages available for biologists. This project will reduce the complexity of obtaining quantitative data from video by developing commercial-off-the-shelf equipment packages, automated video processing software, and the advisory services needed to make this transformational technology more accessible to biologists in the field. The video processing needed is the focus of the research; the goal is the create algorithms that automate object detection and classification from the video and allow non-experts to create an accurate machine learning classifier with minimal labeling requirements. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.