Answers are grounded in this paper and cite their sources.
Seed vigor plays a crucial role in crop quality influencing the growth and yield. The standard germination and accelerated ageing tests are traditional vigor assessment methods that require extended time periods to complete and destroy test samples. The proposed DeiTrix-V, a Vision Transformer-based model utilizes both spatial and temporal representations to assess maize seed vitality through RGB based time series image analysis. The study employs a publicly available seed vigor dataset that encompasses five distinct germination stages: ungerminated, germinating, germinated, primary root and secondary root development. To improve data efficiency, the model incorporates interactive multi-head attention and temporal feature aggregation for robust learning of fine-grained morphological dynamics inherent to the germination process, consequently facilitating accurate multi-class prediction. The experimental results demonstrated that DeiTrix-V achieves superior performance, attaining 98.30% classification accuracy while maintaining high precision, recall and F1-score across all germination stages. A comprehensive evaluation of DeiTrix-V was done using quantitative metrics and visual representations. Further, experiments were conducted to assess the performance of DeiTrix-V in comparison with existing machine learning, deep learning and transformer-based models. The findings demonstrated rapid and non-destructive seed vigor assessment as a strong applicability in seed quality control and precision agriculture.
No takes yet. Share an insight, caveat, or question.
Pushpa et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: