In recent years interest in AI-based crop yield prediction has grown with the adoption of precision agriculture. Image-based approaches can reduce time and cost, but their performance often degrades under varying regional conditions, illumination, and...
In recent years interest in AI-based crop yield prediction has grown with the adoption of precision agriculture. Image-based approaches can reduce time and cost, but their performance often degrades under varying regional conditions, illumination, and crop growth status, limiting generalization.
This study investigated onion and garlic grown from transplanting to harvest at a pilot field of the Muan County Allium Vegetable Research Institute. Multispectral images(400–1,000 nm) were collected using a drone, vegetation indices were derived, and their relationships with growth indicators and bulb weight were analyzed. To reflect crop growth characteristics, the season was divided into bulb enlargement and maturation stages, and stage-wise time-series vegetation indices were modeled using a multivariate functional regression approach.
Vegetation indices from specific growth stages showed correlations with bulb weight in both crops, supporting the feasibility of vegetation index and its potential use in agriculture decision support.