The incorporation of computer vision in contemporary agriculture has witnessed significant advancements, particularly in detecting diseases and deficiencies affecting citrus fruit production. This study provides an in-depth comparative analysis of several limitations of the citrus fruit detection system and the cutting-edge machine vision algorithms used for classification. Traditional diagnostic methods are initially reviewed, followed by an elaborate discussion on various image acquisition techniques such as remote sensing, hyperspectral imaging, bio speckle laser imaging, and color imaging. These techniques focus on extracting features like color, texture, and size for diagnosing citrus fruit diseases. Despite their effectiveness, the images obtained might contain noise and distortions. The study details two crucial steps—image preprocessing and segmentation—to minimize these anomalies. It further explores a range of classification techniques and their efficacy in different research contexts. The paper is structured around five key components: diverse image capture methods, preprocessing and segmentation techniques, various extracted features, classification techniques for citrus fruit detection, and a comparison among classification methods like machine learning, deep learning, and statistical techniques. The study concludes by discussing current challenges and limitations in detecting citrus fruit diseases. It emphasizes the use of thresholding in hyperspectral imaging and identifies RGB color space as a frequently used feature. Among the compared techniques, Support Vector Machine (SVM) in machine learning, Artificial Neural Network (ANN) in neural networks, Convolutional Neural Network (CNN) in deep learning, and Linear Discriminant Analysis in statistical approaches emerge as the most effective.