Sign in Subscribe
Cable television broadcasting industry

Artificial Intelligence Researchers Develop Deep Learning-Based Image Processing Pipeline for Wild Animal Consumption Studies

Strategic Insights

— 1 min read

Artificial intelligence researchers from the University of Georgia have published a study on a deep learning-based image processing pipeline to estimate corn consumption by wild animals. The study aims to test the hypothesis that wild animals prefer non-genetically modified (GMO) corn and avoid eating GMO corn. The research team developed a pipeline that uses a mask regional convolutional neural network (Mask R-CNN) to segment and identify corn and its bare cob from images of consumed corn ears.

Key Takeaways:

  • The researchers developed a deep learning-based image processing pipeline to estimate corn consumption by wild animals, which aims to test the hypothesis that wild animals prefer non-GMO corn and avoid eating GMO corn.
  • The pipeline uses a mask regional convolutional neural network (Mask R-CNN) for instance segmentation and was trained to identify corn and its bare cob from images of consumed corn ears.
  • The research team tested two approaches for segmentation: identifying whole corn ears and bare cob parts with and without corn kernels, and found that the latter approach resulted in superior segmentation performance and estimation accuracy.
  • The Mask R-CNN model was found to provide highly accurate results for estimating corn consumption, with an R2 value of 0.99 compared to manually labeled test data.
  • The research has potential applications in plant phenotyping tasks, such as yield estimation and plant stress quantification.

Statistics:

  • The research team used an R2 value of 0.99 to evaluate the accuracy of the Mask R-CNN model in estimating corn consumption.
  • The study found that the model performed better when identifying bare cob parts without corn kernels, with a segmentation performance of 99%.
  • The research team trained the Mask R-CNN model on a dataset of 1000 images of consumed corn ears.

Sources:

  • Frontiers in Artificial Intelligence, 2021,3: Instance Segmentation to Estimate Consumption of Corn Ears by Wild Animals for GMO Preference Tests.
  • doi-org.sdpl.idm.oclc.org/10.3389/frai.2020.593622
  • Institute of Artificial Intelligence, University of Georgia, Athens, GA, United States
  • University of Georgia, Athens, Georgia, United States
  • NewsRx LLC, Robotics & Machine Learning, February 15, 2021; p 578.

Read more