Saratov Region Develops Neural Network to Identify Sterile Sunflower Plants
The technology could help breeders develop hybrid seed varieties.
Members of the interuniversity startup AgroWeed have begun the field phase of data collection for a new neural network. The model will be able to distinguish between fertile and sterile sunflower plants, helping breeders develop hybrid seed varieties.
The team collected 1,000 images in the Saratov Region for the project. The images are now being processed and labeled for inclusion in the dataset.
Key features include identifying a plant as fertile or sterile within 2-3 seconds and precisely outlining it on screen; processing data directly on the device when no network connection is available; and sending images to a server equipped with powerful GPUs for more accurate analysis when internet access is available. The model will also continuously learn from images collected by the developers.
