The growing global demand for Agaricus bisporus has focused on automated harvesting systems, prompting the adoption of artificial intelligence to enhance precision and efficiency. This study aimed to prove the possibility of automated analysis for mushroom phenotypic traits including pileus diameter and color parameters (L*, a*, b*) by using AI model, YOLOv11-seg. Mushroom images were obtained in custom-designed imaging chamber and image training was processed using YOLOv11-seg. By achieving an mAP50 of 0.96, model demonstrated high detection and segmentation performance with stable predictive behavior. To evaluate biological validity, predicted phenotypic traits were compared with mechanically measured values. Pearson correlation coefficient analysis showed that the correlation coefficient for chromaticity was above 0.69, while the correlation coefficient for shoulder diameter was very low at 0.03. Linear regression analysis showed correlations above 0.69 for all phenotypic traits, indicating that the model analysis reflected the actual measurement variation well. Mean absolute error (MAE) analysis showed less than 10% error of 1.32, 2.43, 0.55, and 0.90 in pileus diameter, L*, a*, and b*, respectively, resulting in significant model accuracy. Based on these results, YOLO-based estimation of pileus area was processed to prove the model’s capacity to extract phenotypic traits beyond the limits of traditional analysis. These results indicate that AI models including YOLOv11 show the possibility of the automated growth monitoring for the next-generation smart cultivation systems.

NGS data was yielded by using Illumina Hiseq platform. The short reads were filtered by quality score and read length were mapped against the reference genome (KACC42780). Genome-wide reanalyzed data of Flammulina strains were compared against the reference genome to establish a genome-wide single nucleotide polymorphism (SNP). The rate of mapping differences between the strains reflected in the strain variation in its result. The genome-wide SNPs distribution divided into types of homozygous SNP and heterozygous SNP moreover all of the strains demonstrated a wide variation in all of the regions. In the further study of topological relationship between the collected strains, phylogenetic tree was separated into 3 major groups. Group I contained F. velutipes var. related strains of ASI 4062, 4148, 4195. Group Ⅱ contained strains that were different species of ASI 4188 F. elastica, ASI 4190 F. fennae, and ASI 4194 F. rossica. The other 19 strains F. velutipes were classified as a single group. Polymorphic SNPs of F. velutipes strains representing the phylogenetic segregation of whiteand brown-fruiting body forming groups were compared. As previously reported, white gene expression is recessive to brown in fruiting body color gene expression. The white strains produced 131,874 SNPs to be aa type and homozygous from of SNP. 407,947 SNPs were detected as AA, Aa type from the brown-fruiting body of SNP. We constructed a SNP matrix with 8 white strains and 12 brown strains. To develop the molecular marker related in to fruiting body color and geographical origin, we isolated 240 SNPs from the white-and brown-fruiting body forming. To determine the chromosome relationship on polymorphic SNP between Korea and Japan strains producing white-fruiting body, we analyzed that the Korea white strains detected 185,695 SNPs and the Japan white strains produced 263,811 SNPs. Using the constructed SNP matrix with 3 Korea white strains and 3 Japan white strains, the experiment generated 475 SNPs of phylogenetic SNPs fromKorea and Japan white-fruiting body. As a result, we regarded as they are potentially related to the white color. White and brown color and origin specific SNPs could be used as an identification marker for selection of F. veluipes strains in the breeding program.