Abstract Okra (Abelmoschus esculentus L.), an economically important vegetable in South Gujarat which faces reduced productivity due to multiple foliar diseases effected by fungal, bacterial, viral and pest-associated pathogens. Large scale field deployment is hindered by the labor- intensive and subjective characteristic of traditional disease detection. Automated image-based diagnosis is now possible by current developments in machine learning and deep learning however standalone models are frequently unreliable in real-world scenarios because of interclass similarity, backdrop complexity, illumination variations and a lack of labelled data. This systematic review studies hybrid ML and DL approaches for plant leaf disease detection which is focusing on okra and South Gujarat agro-climatic conditions. Articles published between 2019 and 2026 were analyzed which is covering hybrid frameworks that integrate CNNs, vision transformers, handcrafted feature extraction, classical ML classifiers, ensemble learning and segmentation/localization models. The review estimate datasets, real-time applicability, model, feature fusion and performance measures. Output indicates that hybrid frameworks perform better than single models in call of accuracy and generalization however, there are still shortcomings including a dearth of okra datasets relevant to a given location, a dearth of early-stage and multi-disease detection studies, poor interpretability and a weak interaction with precision agriculture systems.