Researchers at the University of California, Davis, have developed a groundbreaking approach to track and manage the invasive weed johnsongrass using artificial intelligence (AI) and machine learning. Leveraging Google's Street View database, the team created "Google Weed View," a cost-effective and efficient tool that identifies and locates johnsongrass in the Western United States.

Key Points:

1. Data Source: The researchers utilized Google Street View's vast photo database of roadways, streets, and highways to identify johnsongrass locations.

2. Algorithm Training: The AI model was trained using labeled images of johnsongrass, initially focusing on Texas where the weed is prevalent. The deep learning model iteratively improved as it processed more images.

3. Cost and Efficiency: Google Weed View demonstrated significant cost savings, with the entire process costing less than $2,000. In contrast, a traditional in-person survey covering the same area would incur an estimated $40,000 in expenses.

4. Scalability: The AI model can be easily extended to track other plant species by labeling new items in Street View photos and training the algorithm to identify those objects.

5. Applications: The tool provides a rapid and convenient method for surveying invasive plants, offering potential applications in land management and ecological research.

6. Climate Impact: Google Weed View, by providing location information, enables the examination of how climate influences the growth and spread of weeds and invasive plants on a large scale.

Significance:

• Google Weed View offers a transformative approach to weed tracking, providing scalability, cost-effectiveness, and efficiency compared to traditional survey methods.

• The tool's success highlights the potential of AI and machine learning in ecological research and environmental monitoring, paving the way for broader applications in vegetation management.

• The use of Google Street View as a data source demonstrates the innovative integration of existing technologies for addressing real-world challenges in agriculture and ecology.

More: https://phys.org/news/2023-12-algorithm-ai-enables-low-cost-tracking.html

You may be interested in:

Is obesity a disease? Not always, new expert report says

A team of more than 50 doctors and scientists has concluded that obesity is not always a disease. Their report today in The Lancet Diabetes and Endocrinology, which presses for a more nuanced…