Artificial Intelligence Identification of Japonica Rice varieties Based on Raman Spectroscopic Identification Mechanism of Saline-alkali Tolerance

Abstract

Abstract Rice is regarded as the preferred crop for saline-alkali soil improvement by researchers. At present, the identification method for saline-alkali tolerance of rice varieties requires researcher to conduct tedious field investigations based on growth indicators. Therefore, there is an urgent need for an effective technical means to quickly and accurately identify saline-alkali tolerance of rice varieties. Study used 20 japonica rice varieties with three types of saline-alkali tolerance as test materials, by analyzing the identification mechanism of salt-alkali tolerance in Raman spectrum of japonica rice varieties, seven characteristic spectral peaks closely related to salt-alkali tolerance were identified. Various algorithms in Python are used for data standardization, baseline elimination, extraction of characteristic spectral peaks, detection of peaks characteristic information and data noise reduction. Three identification models were established to confirm the highest accuracy of CapsNets identification model, which could provide technical support and reference for breeding saline-alkali resistant japonica rice varieties.

Plant phenotyping relevance

ラマン分光とスペクトル処理・AIモデルによりイネ品種の塩類アルカリ耐性を推定する技術を開発・比較しており、表現型状態の取得・判定が研究の中心である。

Code and data availability

The paper's Data Availability statement explicitly states that the data and code (Raman spectral phenotyping data and Python analysis/identification models for saline-alkali tolerant japonica rice) are openly available in the authors' public GitHub repository.