Journal of Basic and Applied Research International

Plant pests and diseases remain among the most persistent threats to global food security, reducing potential crop yields substantially every year and imposing heavy economic losses on producers across both smallholder and industrial farming systems. Conventional surveillance relies on manual scouting and visual diagnosis, approaches that are slow, subjective, and poorly suited to the scale and speed at which modern outbreaks develop. Over the past decade, artificial intelligence and machine learning have reshaped the surveillance landscape, offering tools that span image-based detection, sensor-driven monitoring, predictive modelling, and decision support. This review synthesises developments across this continuum, tracing the progression from convolutional neural networks and vision transformers used for leaf-level disease classification, through object detection frameworks and Internet of Things-enabled smart traps used for insect pest monitoring, to remote sensing and species distribution models used for landscape-scale forecasting, and finally to decision support systems that translate model outputs into actionable field guidance. Particular attention is given to persistent technical constraints, including the scarcity and imbalance of labelled field data, poor generalisation of models trained on curated datasets when deployed under variable field conditions, and the limited interpretability of deep architectures, alongside emerging responses such as transfer learning, generative data augmentation, and explainable artificial intelligence. The review further considers infrastructural and adoption barriers relevant to resource-constrained farming systems, including connectivity, computational cost, and the integration of artificial intelligence outputs into existing extension and decision-making structures. Drawing on this synthesis, priority directions for future research are identified, including federated and privacy-preserving learning across distributed agricultural datasets, multimodal sensor fusion, and closer coupling between predictive analytics and site-specific intervention. The findings indicate that while detection accuracy has advanced considerably, the translation of artificial intelligence outputs into reliable, trusted, and economically viable decision support remains the central unresolved challenge for the field.