Artificial Intelligence Based Crop Disease Detection and Agricultural Productivity with Environmental Improvement Using Machine Learning

Abstract

Artificial intelligence (AI) and machine learning are increasingly used to identify visible crop disease symptoms and support timely crop-management decisions. Agricultural productivity or environmental sustainability, however, does not follow naturally from high image-classification performance. A critical narrative review was undertaken to explore the relationship between diagnostic reliability, field deployment, agronomic decision making, farmer adoption, and environmental outcomes. It reflects twenty selected publications about image classification and segmentation, smartphone diagnosis, precision spraying, adoption, pesticide management, and environmental accounting from 2016 to 2026. Only targeted searching and purposive selection were used; it is not a systematic review or meta-analysis, and no experiments or pooled effect estimates are reported.

High accuracy on controlled image datasets can be reduced in field conditions due to changing lighting and backgrounds, crop cultivars, symptoms, and sampling. Designs on mobile diagnostic systems and recent crowd-management studies address real opportunities but sometimes no longer separate the extra value of AI from treatment, training or extension support. Similarly, reports of reductions in pesticide application due to better management practices are not necessarily follow-on effects of disease-detection technologies. Hence a framework of representative field observations, uncertainty assessment, agronomic verification, feasible farmer adoption, targeted management, and independent measurement of outcomes is proposed by the present review. Pesticide risk, resource use, and equitable access shall be considered at the same time as diagnostic performance, marketable yield and farm profitability.