科研速览 · Science Skim继续刷下去 · Keep skimming →
◆ Food chemistry2026-09-03

Volatile profile evolution and simultaneous identification of pumpkin cultivars and postharvest stages using a multi-scale multi-label attention network.

Yingchao Xu, Jiayu Luo, Huihui Han, Shudan Xue, Manman Wang, Wenlong Luo, Qingmin Jin, Weijian Zhou, Yujuan Zhong

原始摘要(英文原文)· Original abstract
Pumpkin exhibits excellent postharvest storability; however, cultivar-specific changes in volatile profiles during storage remain poorly understood. The total volatile compound content increased progressively in Cmo-gm, Cmo-53, and Cma-81 but increased initially and then declined in Cma-78. Aldehydes dominated volatile profile changes in C. maxima, whereas alcohols predominated in C. moschata, accompanied by continuous geraniol accumulation. Based on volatile and nonvolatile quality profiles, favorable postharvest periods were identified at 20-30 dop for C. moschata and 10-20 dop for C. maxima. A multi-scale, multi-label attention network was developed using Fourier transform near-infrared spectroscopy for rapid cultivar and postharvest stage identification, achieving a Hamming loss of 0.0065, a macro F1-score of 0.984, a micro F1-score of 0.985, and a subset accuracy of 0.956. These findings reveal postharvest changes in volatile evolution profiles and support rapid cultivar and postharvest stage identification under controlled storage conditions.
读原文 · Read the paper ↗

AI 追问PRO

登录后使用 AI 追问

讨论区

登录后参与讨论

相关论文 · Related

Volatile profile evolution and simultaneous identification of pumpkin cultivars and postharvest stages using a multi-scale multi-label attention network. — 科研速览 Science Skim