Sunflowers are an important cash crop, but the appearance of sunflower volunteers in surrounding crops can cause significant losses. Sunflower volunteers act as weeds and compete with other crops for nutrients, water, and sunlight. To prevent losses, sunflower volunteers must be removed in a timely manner.

The importance of removing sunflower volunteers cannot be overstated. Studies have shown that sunflower volunteers can reduce crop yields in wheat, soybean, and oilseed rape by up to 50% if not removed or if done so too late. There are two main methods for removing sunflower volunteers: manual removal and herbicide use.

Manual removal involves physically pulling out the sunflower volunteers by hand. While this method is effective, it can be time-consuming and costly. Herbicides are another option for removing sunflower volunteers. However, there are concerns about the potential for herbicide resistance, and the cost of herbicides can also be a significant expense for farmers.

Recently, a new hybrid approach has emerged that significantly improves upon the traditional manual removal methods, drones and machine learning algorithms could be used to identify sunflower volunteers in crops, making removal much more efficient.

Not only that drones can help with the detection of sunfloweer volunteers but they can also help with early detection which is very important because it allows farmers to remove sunflower volunteers before they have a chance to compete with other crops for resources. By using drones to scan their crops, farmers can detect sunflower volunteers earlier and remove them before they cause significant damage.

Drones equipped with high-resolution cameras can capture images of the field, which are processed into a map and then analyzed using machine learning algorithms to detect the presence of sunflower volunteers. The algorithms used for this task are trained on large datasets of annotated maps, allowing them them to learn to identify sunflower volunteers based on their visual characteristics.

One of the advantages of using machine learning algorithms for sunflower volunteer detection is their ability to adapt and improve over time. As more data is captured and analyzed, the algorithms can learn to recognize new patterns and variations of sunflower volunteers, improving their accuracy and reducing false positives.

In addition to detecting sunflower volunteers, these algorithms can also provide other useful information to farmers, such as the severity and distribution of sunflower volunteer infestations across the field. This information can help farmers make informed decisions about where to focus their manual removal efforts and where to apply herbicides or other treatments.

The use of drones and machine learning algorithms for sunflower volunteer detection is already being successfully implemented in some areas. For instance, Corteva Agriscience alongside Skyline Drones has used this technology to identify sunflower volunteers in over 6,000 hectares of crop fields, achieving significant improvements in the efficiency and accuracy of sunflower volunteer removal.

For our sunflower volunteer detection project, we focused on six areas where sunflower volunteers have been known to cause significant losses in neighboring crops. To ensure the accuracy of our detection methods, we added a 1.5 km buffer zone around each field, from which populated or constructed areas, forests, and water bodies were removed. This approach allowed us to focus specifically on the crop fields and the surrounding areas where sunflower volunteers are most likely to occur, increasing the efficiency and accuracy of our detection efforts.

To cover the six areas of interest, we used the Trinity F90+, a high efficeincy drone, to scan each zone. Each of these six zones were quite large, so we divided them into smaller chunks of 4-6 sections to ensure thorough coverage. This approach allowed us to scan each section in detail, providing us with high-resolution images and adequate GSD of around around 2 cm/px. The processed maps were then analyzed using machine learning algorithms to detect the presence of sunflower volunteers. By using this approach, we were able to achieve a high degree of accuracy in our detection efforts, allowing us to pinpoint the locations of sunflower volunteers with greater precision.

To optimize the effectiveness of our drone scanning efforts, we conducted the scans during the R1 to R2 stages of sunflower development. This is because sunflower volunteers can be easier to detect during these stages, as they are still growing and have not yet fully developed their characteristic yellow petals. By scanning the fields during these stages, we were able to identify sunflower volunteers before they could cause significant losses in neighboring crops. This allowed farmers to take action quickly to remove the sunflower volunteers, preventing them from spreading further and reducing the overall economic impact of their presence.

We conducted our drone scanning efforts to detect sunflower volunteers in the six areas of interest at the end of June, which took a total of 7 working days. During this time, we captured a total of 80,000 images of the crop fields and surrounding areas using our drone. These images were then processed using machine learning algorithms to identify the presence of sunflower volunteers. The large number of images captured during this effort allowed us to thoroughly cover the areas of interest and achieve a high degree of accuracy in our detection efforts.

While the technology is still in its early stages, its success in the field demonstrates its great promise for improving seed purity and reducing losses caused by sunflower volunteers. As the technology continues to develop, it is likely that it will become an increasingly important tool for farmers in the fight against sunflower volunteers.

For more details, contact us at phone number +40724 339 757, at the e-mail address ciprian.iorga@laorizont.ro or through this form.