Abstract: Fruit tree pruning is an important horticultural practice for regulating tree structure, improving canopy ventilation and light conditions, and stabilizing fruit yield and quality. Traditional pruning mainly depends on manual operation and practical experience. Differences in workers’ understanding and execution of pruning standards can lead to missed or incorrect cuts, while labor shortages and rising costs further limit manual pruning. With the development of computer vision, deep learning, and three-dimensional sensing, fruit tree information acquisition has expanded from two-dimensional appearance description to three-dimensional structural representation, providing a basis for transforming manual observation and empirical judgement into machine-processable information. Existing reviews have mainly focused on general plant point-cloud segmentation or robotic pruning hardware, while systematic analyses of the complete process from visual perception and pruning decision-making to robotic execution have remained insufficient. Therefore, this study reviewed recent progress in computer-vision-based intelligent fruit tree pruning from three interconnected levels. At the perception level, the main methods for acquiring fruit tree images and three-dimensional information were summarized, together with image and point-cloud preprocessing methods. Existing studies improved tree reconstruction and structural information acquisition, but no single sensing scheme was found to be suitable for all pruning scenarios. Fine pruning of individual trees required complete representation of small branches and topological relationships, whereas continuous online operation required a balance among modeling accuracy, acquisition efficiency, and equipment cost. At the decision-making level, image- and point-cloud-based methods for branch segmentation and recognition, structural parameter extraction, pruning branch identification, and pruning point localization were reviewed. Deep learning improved branch recognition and segmentation, while three-dimensional data provided quantitative information on branch length, angle, spacing, diameter, and topology. However, category-level segmentation alone was insufficient when branches intersected or adhered and pruning targets had to be identified at the individual-branch level. Pruning decisions therefore needed to combine branch instance separation, three-dimensional structural parameters, and horticultural pruning rules. The lack of large public datasets and the difficulty of quantifying experience-based horticultural knowledge still restricted model generalization and decision accuracy. At the execution level, hand-eye coordinate transformation, robotic arm path planning, and pruning end-effector design were summarized. Existing studies demonstrated the feasibility of converting pruning targets into robot-executable positions and motions, but depth measurement, calibration, and kinematic errors could accumulate during coordinate transformation. Static or simplified obstacle models were also insufficient to account for wind disturbance and flexible branch deformation, while current end effectors showed limited adaptability to different branch conditions. Overall, computer vision established a technical chain from tree information perception to pruning decision and robotic cutting. The key challenge was not simply to maximize the accuracy of an individual perception, segmentation, or planning algorithm, but to ensure that tree structures obtained at the front end could be correctly interpreted by horticultural rules and converted into pruning points and cutting poses that robots could execute accurately. Future research should strengthen multi-source information fusion and adaptive preprocessing, improve model generalization and the computability of horticultural knowledge, use online sensing to update pruning points and motion paths in dynamic canopies, and develop more accurate, reliable, and practical integrated pruning systems, thereby promoting intelligent fruit tree pruning from experimental research toward practical orchard application.