Global Wheat Full Semantic Segmentation (GWFSS)

WHEAT is a cornerstone crop of global food security, serving as a dietary staple for billions of people worldwide. Detailed analysis of wheat plants can help plant scientists and farmers grow healthier, more resilient, more nutritious, and more abundant wheat crops. Analyzing images of wheat in-the-field is a challenging task, due to dense, overlapping, self-similar plant structures and a highly variable appearance due to the type of wheat, the growing region, and the growth stage.

To tackle this global wheat challenge, we previously assembled an international dataset of wheat images focusing on wheat head bounding-box annotations. Although counting wheat heads is an important trait, the other parts of the wheat plant (leaves, stems, etc.) are also important for a holistic characterization of a wheat plant’s condition and potential. Our new global wheat dataset and competition takes a significant step forward by targeting full semantic segmentation of wheat organs. Semantic segmentation enables pixel-level classification of plant components such as leaves, stems, and spikes. This detailed understanding can provide deeper insights into plant architecture, health, and development.

The Global Wheat Full Semantic Segmentation (GWFSS) dataset comprises images collected by 11 institutes and universities from diverse geographical regions using various imaging setups. This diversity ensures that the dataset is robust to variations in environmental conditions, genotypes, and imaging techniques.

By participating in this competition, researchers and practitioners have the opportunity to develop cutting-edge models that can accurately segment wheat organs under diverse conditions. The outcomes of this challenge have the potential to revolutionize digital phenotyping in agriculture by enabling more precise monitoring of crop health and growth. Ultimately, these advancements can contribute to global efforts in improving food security and fostering sustainable agricultural practices. This structure highlights the significance of the task while connecting it to broader goals like food security and sustainability. It also emphasizes the novelty of transitioning from detection to segmentation and underscores the importance of the new dataset.

Two travel prizes will be awarded for the GWFSS Competition to attend the workshop:

  1. Top Performance Award.
  2. Innovation Award.

The travel award will consist of reimbursement of flight, accommodation, and conference registration fees (up to a maximum of $4,000 USD) for presenting the competition solution at CVPPA. For more details please click here!

Participate Now!

Paper contribution should be submitted in accordance with the Call for Paper