iRSVPred

iRSVPred identifies basmati rice seed varieties using deep learning on seed images to enable rapid and accurate variety differentiation for agricultural applications.

Key Features:

• Deep Learning-Based Identification: Utilizes deep learning algorithms to analyze seed images for the identification of basmati rice varieties.
• High Accuracy: Achieves 100% accuracy on the training set and 97% on the internal validation set.
• Comprehensive Coverage: Identifies ten basmati rice varieties: Pusa Basmati 1121, Pusa Basmati 1509, Pusa Basmati 1637, CSR 30 (salt-tolerant), Dehradoon Basmati Type-3, Pusa Basmati-1, Pusa Basmati-6, Basmati -370, Pusa Basmati 1718, and Pusa Basmati 1728.
• External Validation: Reports accuracies of 80% or higher for all varieties on an external validation dataset.

Scientific Applications:

• Agricultural Enhancement: Facilitates identification of high-yielding and quality paddy seeds to support crop yield optimization.
• Alternative to Molecular Techniques: Provides an image-based alternative to laboratory molecular assays for seed variety identification, reducing reliance on laboratory resources.

Methodology:

Deep learning models were trained on 61,632 seed images, with internal validation on 15,408 images and external validation on 642 images.