One-Time Training for All Grains: Open-Set Grain Recognition and Quantitative Analysis

2026-08-10

AI Summary

Traditional grain recognition methods require re-collecting data and retraining models when new varieties are introduced, hindering efficient scalability. This work proposes GROW, a novel open-set recognition framework that, for the first time, enables continuous integration of new grain varieties after a single training phase. GROW employs category-agnostic localization to extract individual grains, fuses visual embeddings with morphological descriptors to construct an extensible GrainBank, and leverages a weighted top-k retrieval mechanism to achieve recognition, counting, and phenotypic analysis without model retraining. The method demonstrates robust performance under varying conditions—including variety expansion, density changes, and background shifts—reducing new variety registration time from 4,153 seconds to just 39 seconds while maintaining high identification accuracy.

Abstract

Advances in crop breeding have introduced an increasing number of grain varieties, creating a growing demand for efficient variety recognition and quantitative analysis. However, existing methods are typically trained on a fixed variety set, and incorporating newly introduced varieties requires additional data collection and model retraining. To address this limitation, we propose GROW, a framework for Grain Recognition and quantitative analysis in Open sets Without retraining. GROW first performs class-agnostic grain localization, converting mixed-grain images into individual instances for variety-wise counting and phenotypic measurement. It then combines visual embeddings and morphological descriptors into fused grain descriptors stored in an extensible GrainBank. Query grains are recognized through rank-similarity weighted top-k retrieval, and newly introduced varieties are incorporated by appending their descriptors without updating the deployed models. Extensive experiments under progressive variety expansion, varying grain densities, and background domain shifts demonstrate the scalability, robustness, and adaptability of GROW. Compared with joint retraining, GROW reduced the average category-registration time from 4153 s to only 39 s while maintaining competitive recognition performance. These results demonstrate that GROW provides an efficient and maintainable solution for extensible grain recognition, counting, and phenotypic analysis without repeated model retraining.

Problem

• open-set recognition
• grain variety identification
• quantitative analysis
• model retraining
• scalable recognition

Innovation

• open-set recognition
• zero-shot learning
• grain phenotyping
• extensible descriptor bank
• retraining-free adaptation