Skip to main content
. 2026 Feb 10;17:1747657. doi: 10.3389/fpls.2026.1747657

Figure 4.

Three sets of aerial images in rows show farmland with green crops and soil. Each row progresses from an original image on the left to processed images with increasing red square markings highlighting specific areas on the crops, from (b) to (e). The squares represent detected items or features within the crops.

Segmentation results of three different models: (a) original image; (b) FCN; (c) DeepLabv3+; (d) U-Net; (e) SE-UNet.