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

Figure 3.

This figure demonstrates a comparative analysis of aerial imagery of farmland processed through various semantic segmentation models. From left to right, the images represent: the original image, FCN, DeeplabV3+, UNet, and SE-UNet segmentation results. The red boxes highlight areas of missed and redundant detections within each model's segmentation output, providing a visual comparison of algorithmic performance in crop identification tasks.

Data collection and processing, model construction and analysis flowchart.