Location

Online

Event Website

https://hicss.hawaii.edu/

Start Date

4-1-2021 12:00 AM

End Date

9-1-2021 12:00 AM

Description

To prevent yield losses, it is critical to eliminate competition between food crops and weeds at the onset of plant growth. While uniform spraying of herbicides can be economically and environmentally inefficient, site-specific weed management (SSWM) counteracts this by reducing the amount of chemical application with localized spraying of weed species. Past research on weed detection in SSWM has used a large deep convolutional neural network (DCNN) for weed detection. These models are, however, computationally expensive and prone to overfitting on smaller datasets. In this paper, we propose an approach to detecting weeds amongst plant seedlings using transfer learning in a small network. Our approach combines the mobile-sized EfficientNet with transfer learning to achieve up to 95.44% classification accuracy on plant seedlings. Due to the robustness of transfer learning methods, this approach would be beneficial in improving both the classification accuracy and generalizability of current weed detection methods.

Share

COinS
 
Jan 4th, 12:00 AM Jan 9th, 12:00 AM

An Approach for Weed Detection Using CNNs And Transfer Learning

Online

To prevent yield losses, it is critical to eliminate competition between food crops and weeds at the onset of plant growth. While uniform spraying of herbicides can be economically and environmentally inefficient, site-specific weed management (SSWM) counteracts this by reducing the amount of chemical application with localized spraying of weed species. Past research on weed detection in SSWM has used a large deep convolutional neural network (DCNN) for weed detection. These models are, however, computationally expensive and prone to overfitting on smaller datasets. In this paper, we propose an approach to detecting weeds amongst plant seedlings using transfer learning in a small network. Our approach combines the mobile-sized EfficientNet with transfer learning to achieve up to 95.44% classification accuracy on plant seedlings. Due to the robustness of transfer learning methods, this approach would be beneficial in improving both the classification accuracy and generalizability of current weed detection methods.

https://aisel.aisnet.org/hicss-54/da/analytics_for_green_is/4