Extraction Study of Leaf Area and Plant Height of Radish Seedlings Based on SAM

Leaf area and plant height are important indicators of plant growth. This study introduced an innovative method for automating the measurement of leaf area and plant height in radish seedlings. Using the Segment Anything Model (SAM), which comprises an image encoder, prompt encoder, and mask decoder, images of radish seedlings were captured and analyzed. By comparing labeled images with SAM-generated segmented images, key evaluation metrics, such as MIoU and MPA, were used for pixel-level segmentation assessment. The experimental results demonstrated an impressive MIoU of 88.83%, outperforming traditional methods and even improving the MPA by 2.8% when compared to the U-net segmentation network. This intelligent and nondestructive approach offers precise data extraction, providing an efficient way to monitor and enhance radish seedling yields.