Use of Synthetic Images for Training a Deep Learning Model for Weed Detection and Biomass Estimation in Cotton

This study investigates the use of synthetic images to train a convolutional neural network (Mask R-CNN) for weed detection and segmentation in cotton fields. Synthetic images were generated using plant instances clipped from UAV-borne real images and Generative Adversarial Networks (GANs). The results show that models trained on synthetic images, combined with a small amount of real data, achieved comparable or better performance than those trained solely on real images, particularly when real data is limited. The approach addresses the challenge of requiring large labeled datasets for training. To read the full article click on the link... https://www.nature.com/articles/s41598-022-23399-z