In a study published in Smart Agricultural Technology, a research group led by ZHAO Qinglei from the Changchun Institute of Optics, Fine Mechanics and Physics (CIOMP) of the Chinese Academy of Sciences, together with collaborators, developed a deep learning method for screening high-vigor maize seeds before germination using ordinary RGB images. The method, called SeedFocusNet, achieved 74.39% precision in identifying high-vigor seeds, while greenhouse tests showed that seeds selected by the model had emergence rates 18–22 percentage points higher than unscreened seeds.

Seed vigor reflects a seed's ability to germinate rapidly and develop into a healthy seedling, making it an important indicator for breeding and pre-sowing selection. Conventional vigor tests can be destructive and may require several days of labor. Techniques such as X-ray CT and hyperspectral imaging can provide detailed information but rely on specialized equipment. Ordinary RGB cameras are much cheaper and easier to use, yet vigor is mainly an internal physiological property, while photographs reveal only subtle differences in seed color, texture, and shape. This makes early vigor prediction from a single image particularly challenging.

To investigate whether these subtle visual clues could support practical screening, the researchers collected pre-germination RGB images of 955 maize seeds and then conducted controlled germination tests to determine their actual vigor. They processed the images to remove background interference and extracted informative regions such as the seed ends, hilum, and possible surface damage. These features were analyzed by SeedFocusNet, which used Focal Loss to place greater attention on difficult-to-distinguish seeds. Bayesian optimization was further used to automatically search for suitable model parameters, reducing reliance on manual trial and error.

The researchers compared several commonly used deep-learning architectures and found that ResNet50 provided the best overall balance for practical screening. On the independent test set, SeedFocusNet achieved 74.39% precision and 54.46% recall, compared with 67.53% precision and 46.43% recall for an optimized conventional ResNet50 baseline. Greenhouse experiments provided further validation: seeds selected by SeedFocusNet achieved emergence rates of 78%, 76%, and 68% after one, two, and three years of storage, respectively, compared with 60%, 54%, and 48% for unscreened seeds.

The study demonstrates the potential of combining inexpensive RGB imaging with deep learning for early seed-vigor screening. By identifying promising seeds before planting, the approach could provide breeders and seed producers with a more accessible alternative to slower or equipment-intensive testing methods. The researchers noted that the current system was evaluated on one maize accession under controlled laboratory conditions, and further testing across different varieties, seed batches, imaging environments, and field conditions will be needed before large-scale practical application.