MothNet

MothNet is an advanced AI-driven tool developed by Penn State University to assist gardeners, nurseries, researchers, and citizen scientists in accurately identifying the invasive Box Tree Moth. By utilizing cutting-edge computer vision technology, MothNet can differentiate between the Box Tree Moth and other similar-looking native moth species. Inspired by an emerging need identified by Penn State Extension, this tool is critical for tracking the spread of this invasive pest and aiding in effective management strategies.

MothNet is an ongoing research project managed by Penn State INSECT NET graduate trainee Yanqiu Yang. Ms. Yang is a PhD candidate in the Department of Agricultural and Biological Engineering and is advised by Dr. Paul Heinemann (ABE), Dr. Christina Grozinger (Entomology) and Dr. Harland Patch (Entomology). Learn more about Yanqiu by visiting her directory page.

Read on to learn more and to test out the MothNet algorithm on your own!

How to Use MothNet:

Any user can submit images of a suspected Box Tree Moth to the site for species confirmation.

Begin by visiting the MothNet website and creating/logging in to an account.

Next, upload your suspected Box Tree Moth photo from you local device by clicking 'Choose File'

After choosing your image, click “Identify.” The AI model will run automatically to identify the moth. Simultaneously, a window will prompt you to input the location of this observation

Input Location. Location data helps our team to map the moth's spread more efficiently and effectively.

You can input the location that the image taken in two ways:

• Manually enter the latitude and longitude
• Query the GIS coordinates of your location (note - this will populate your current location and not the location where your photo was taken, if these vary)

A confidence level above 90% is considered to be highly reliable

More about the project

The Box Tree Moth, originally from eastern Asia, has emerged as a significant invasive pest in North America, first detected in Canada in 2018 and later spreading to several U.S. states (link to Penn State Extension: https://extension.psu.edu/box-tree-moth). MothNet is a key component of a broader initiative to combat this pest. This decision support tool leverages a trained YOLOv8 model for image classification, utilizing data from iNaturalist to accurately identify Box Tree Moths among similar-looking native species.

MothNet integrates user-submitted data, aiding in the mapping and management of the moth's distribution. The model, trained on a comprehensive dataset, ensures high accuracy in identification. Additionally, location data, either manually entered or retrieved via GIS coordinates, allows the tool to efficiently track and predict the moth's spread. Beyond the web portal, our team is also developing a smart trap for automatic monitoring, further enhancing data collection and management efforts.

Funding for MothNet was provided by the Penn State College of Agricultural Sciences Technology for Living Systems Center and the INSECT NET program (NSF Grant #2243979").