Spectral reflectance dataset of maize seed varieties measured using Vis/NIR spectroscopy with an AS7265X multispectral sensor for seed quality classification

This dataset contains multispectral spectral reflectance measurements of maize seeds collected using a portable Visible/Near-Infrared (Vis/NIR) spectroscopy system equipped with an AS7265X multispectral sensor. The sensor records spectral responses across 18 discrete wavelength channels ranging from 410 nm to 940 nm, covering the visible and near-infrared regions relevant to optical characterization of agricultural seeds.

The dataset was developed to support the analysis and classification of maize seed quality using non-destructive optical sensing combined with machine learning techniques. Spectral measurements were obtained from maize seed samples using a portable Vis/NIR spectrometer designed for rapid spectral acquisition under controlled conditions.

Each dataset entry consists of 18 spectral reflectance intensity values representing the response of maize seeds across the Vis/NIR wavelength range. In addition to spectral information, the dataset includes two categorical attributes:
1. Class (Maize Variety Category) - This column represents the maize seed variety or usage category:
- Seed_SR – maize seed from the SR variety
- Seed_K – maize for consumption (non-seed category)
- Seed_MPM – maize seed from the MPM variety
- Seed_Max – maize seed from the MAXXI variety

2. Quality (Seed Quality Category) - Each sample is labeled according to seed quality level:
- Super
- Premium
- Fair
- Poor

These categories represent different levels of maize seed physiological quality and are associated with important seed performance indicators such as moisture content, germination rate, and electrical conductivity. Spectral measurements were performed using a portable Vis/NIR spectrometer capable of capturing spectral responses in the 410–940 nm range. The measurement configuration included a reflectance chamber designed to minimize ambient light interference and maintain consistent sensor-to-sample geometry. Multiple measurements were performed across different seed positions to ensure representative spectral responses. This dataset provides a valuable resource for research in precision agriculture, seed quality monitoring, spectral analysis, and machine learning-based classification of maize seed quality. The spectral data can support the development of non-destructive seed evaluation systems using affordable portable spectroscopy devices.

Key Features:
- Multispectral reflectance measurements from 18 wavelength channels (410–940 nm)
- Data acquired using portable Vis/NIR spectroscopy with AS7265X sensor
- Four maize seed categories representing variety and seed usage types
- Four seed quality classes: Super, Premium, Fair, Poor
- Non-destructive measurement method suitable for rapid seed quality evaluation
- Applicable for machine learning classification and spectral analysis