Ivan Grijalva, Saurav Upadhyaya, Michael J. Stout and Prashant Jha

Louisiana agriculture contributes approximately $12.96 billion to the U.S. economy each year and is essential for supplying food and fiber nationwide. As agricultural systems face increasing pressure to produce more with fewer resources, pests, weeds and plant diseases continue to threaten productivity, farm profitability and surrounding ecosystems. In response, LSU AgCenter scientists are working to advance precision agriculture technologies to better address emerging pest challenges.

A major component of this effort is improving how pests are detected and monitored in real time across agricultural fields, moving beyond traditional approaches that rely heavily on manual scouting. In these practices, producers and consultants inspect crops and weeds, count pests and estimate damage across only small portions of a field. While these approaches remain important, they are labor intensive, time-consuming and often subjective, making it difficult to capture the full extent of pest infestations as farms become larger and more spatially variable.

Recent technological advances are helping overcome these limitations, particularly through the adoption of Internet of Things (IoT) applications in agriculture, which integrate interconnected sensors, devices and communication networks to continuously collect, transmit and analyze agricultural data in real time (Figure 1). Sensor systems now enable more consistent monitoring across larger spatial (relating to space and distance) and temporal (relating to time) scales. Ground-based sensors, imaging platforms and environmental monitors collect continuous data while drones capture high-resolution imagery across entire fields within minutes.

Satellite imagery provides repeated observations over time, revealing trends that are difficult to detect from the ground. Smartphones and modern farm equipment also contribute real-time data streams. Together, these technologies generate numerical and spatial datasets that provide a complete and more objective picture of crop health and pest pressure. However, collecting large volumes of data alone is not sufficient. The real value emerges when data are transformed into usable and actionable information. This is where artificial intelligence (AI), particularly machine learning and computer vision, becomes essential.

Computer vision algorithms identify patterns within large datasets, linking visual, spectral (relating to a spectrum of colors) or environmental signals to test pest presence or crop stress. Machine learning enables automated image analysis, improving consistency and reducing reliance on subjective visual interpretation. Together, these tools enable rapid processing of large datasets and provide timely decision support for crop and pest management, phenotyping (observing traits) and other agronomic practices.

These analytical approaches are already supporting practical applications developed by LSU AgCenter scientists. Weed detection models identify infested patches within fields, enabling spot treatment rather than whole-field applications, with estimated herbicide reductions of 90% or more (Figure 2). Furthermore, herbicide-resistant weeds in corn, soybean and cotton continue to pose a major challenge, costing farmers millions of dollars in losses annually.

Insect detection models help prioritize where scouting should be focused, increasing efficiency without adding labor demands with close to 85% accuracy using drones and mobile devices, according to a 2025 study by LSU and Kansas State researchers (Reference article: https://doi.org/10.1016/j.mlwa.2025.100711).

Computer vision systems are also used in seed grading to detect and quantify insect- and disease-related damage, protecting market value and accelerating breeding programs with 80% or more accuracy using camera sensors (Figure 3). Across these applications, the goal is to focus management efforts where they are needed most and automate tasks that traditionally rely on visual assessment.

Together, these capabilities enable precision pest management — applying the right control tactic in the right place at the right time. Rather than uniform treatments across entire fields, interventions can be tailored to specific zones identified through sensor data and analytics. Site-specific spraying illustrates this approach, where prescription maps guide targeted applications. This strategy reduces chemical use, lowers input costs and minimizes unintended environmental impacts while maintaining effective and profitable pest control.

Drones and robotic platforms further strengthen these systems by linking monitoring, analysis and action into coordinated workflows. Drones are widely used for data collection and are increasingly capable of carrying out targeted pesticide applications. Emerging robotic and autonomous systems integrate sensors, detection algorithms and treatment mechanisms into unified platforms. Examples include drone-based spot spraying guided by machine-generated maps and robotic units designed to detect and treat localized infestations compared to broadcast applications in the field (Figure 4).

Importantly, these technologies are designed to support — not replace — human decision-making. Outputs from machine learning and computer vision serve as decision support tools, helping producers, consultants and managers allocate time and resources more effectively.

By increasing sampling coverage and reducing subjectivity, these approaches provide faster and more consistent feedback. When problems like pests or weeds that no longer respond to herbicides are found earlier, they can be dealt with before they spread, which lowers risk and helps keep farms profitable.

Overall, integrating pest monitoring, field sampling, sensors, computer vision and machine learning offers a practical path forward for integrated crop and pest management. As agriculture faces more pressure to produce food efficiently and in environmentally friendly ways, decision support tools that improve timing, precision and confidence will be essential for meeting the needs of both producers and the environment.

Ivan Grijalva is an assistant professor of precision pest management and data analytics in the LSU AgCenter Department of Entomology. Saurav Upadhyaya is a graduate student in the Department of Entomology. Michael J. Stout is a professor of rice pest management and executive associate dean of the LSU College of Agriculture. Prashant Jha is a professor of weed management in the AgCenter School of Plant, Environmental and Soil Sciences.

This article appears in the summer 2026 edition of Louisiana Agriculture.

Figure 1. Internet of Things (IoT) applications in agriculture. Illustration by Matt Faust

Figure 2. A Palmer amaranth (Amaranthus palmeri) prescription map generated from high-resolution drone imagery. Illustration by Usman Mohammed

Figure 3. A computer vision model designed to detect and assess damages in soybean seeds. Illustration by Peyton A. Falterman and Kevin Hoffseth

Figure 4. Prescription mapping and site-specific spraying. Illustration by Matt Faust