International Journal of Plant & Soil Science

Cauliflower (Brassica oleracea var. botrytis L.) is a nutritionally important cole vegetable and a high-value component of intensive vegetable production. Its yield and market quality are frequently reduced by bacterial, fungal and oomycete diseases and by insect pests that cause direct feeding injury or facilitate secondary infections. Conventional disease and pest surveillance is largely dependent on visual scouting, expert diagnosis and periodic field observations. These approaches remain indispensable, but they can be labour-intensive, subjective and difficult to scale over large production areas. Artificial intelligence (AI), including machine learning, deep learning, computer vision, image segmentation, object detection, remote sensing and predictive analytics, offers a complementary approach for rapid and repeatable crop-health assessment. This review examines the potential role of AI in cauliflower disease and pest management, covering image acquisition, disease classification, lesion localisation, severity estimation, insect identification, automated counting, forecasting, smartphone applications, UAV-based surveillance and AI-enabled integrated pest management. Cauliflower-specific research has demonstrated promising performance of deep-learning models. The Cauli-Det modified YOLOv8 system reported precision of 93.2%, recall of 82.6% and mean average precision of 91.1% for bacterial soft rot, downy mildew and black rot in its experimental dataset (Uddin et al., 2024). Another recent cauliflower study reported 99.85% classification accuracy using EfficientNet-B3 under its experimental conditions (Malik et al., 2025). Broader plant-disease research also demonstrates strong potential for CNNs and transfer learning (Mohanty et al., 2016; Ferentinos, 2018), while pest-detection research supports automated insect recognition and monitoring (Thenmozhi & Reddy, 2019; Chakrabarty et al., 2026). However, field variability, limited crop-specific datasets, class imbalance, symptom similarity, model explainability, computational requirements and inadequate independent validation remain major constraints. Future research should integrate AI predictions with weather, crop stage, pest abundance, disease severity and economic thresholds so that the technology supports practical, affordable and environmentally responsible IPM in cauliflower.