Quantum Active Learning for Vegetable Plant Phenology Annotation

In the vanguard of agricultural informatics, our study probes the nuanced task of annotating vegetable crop imagery by their phenological progressions. The domain’s intrinsic complexity and diversity present formidable analytical challenges, yet it is indispensable for augmenting agronomic efficacy and fortifying global food security. We propose a pioneering methodology, Quantum Active Learning (QAL), which harnesses quantum computational paradigms to encode and critically appraise the informativeness of unlabeled samples, thus isolating the quintessential ones for annotation. Through the deployment of a meticulously curated dataset of eight vegetable crop images from the PhenoCam network, we implement QAL and juxtapose its performance with Random Sampling (RS) and Classical Active Learning (CAL) via a comprehensive suite of evaluative metrics—accuracy, F1-score, and Cohen’s kappa coefficient—to scrupulously assess QAL’s proficiency within the sphere of vegetable crop phenology annotation. The intricacy and diversity of vegetable crop phenology make this topic challenging, yet it is useful for both agriculture and food security.