Managing weeds through hyperspectral imaging: a new tool for producers

Herbicide-resistant weeds are becoming a major problem for farmers across the Canadian Prairies. These weeds reduce crop yields, increase production costs, and make weed control more difficult.

Traditionally, identifying resistant weeds requires time-consuming lab tests or waiting until visible damage appears after spraying. This delays management decisions and can lead to ineffective herbicide use. This study explored a faster and more efficient solution using advanced imaging technology. Researchers used hyperspectral imaging, a method that captures detailed light information from plants combined with machine learning to detect differences between herbicide-resistant and susceptible weeds. The study focused on two major weed species: kochia and wild oat.

Credit: Keshav Singh

Read the full article in the *Canadian Journal of Plant Science*.

Experiments were conducted in both greenhouse and field conditions across Alberta. The results showed that this technology can identify resistant weeds with a good degree of accuracy, even before visible symptoms appear. In some cases, accuracy exceeded 85–90% after herbicide application and remained strong under real field conditions. If implemented, this early detection capability will allow farmers to respond more quickly and apply targeted weed management strategies. By identifying resistant weeds sooner, producers can reduce unnecessary herbicide use, lower costs, and slow the spread of herbicide resistance.

Overall, the study demonstrates that hyperspectral imaging combined with machine learning offers a promising, non-invasive, and scalable tool for improving weed management and supporting more sustainable agriculture practices.

From “Hyperspectral imaging and machine learning to classify herbicide-resistant and -susceptible weeds in field crops,” Keshav D. Singh, Manoj Natarajan, Prabahar Ravichandran, Breanne D. Tidemann, and Charles M. Geddes, Canadian Journal of Plant Science, 13 May 2026, DOI:10.1139/cjps-2025–0244.