For decades, fruit maturity assessment in the field was challenging for producers, researchers, and food supply agencies. Knowing the maturity stage of the fruit is significant for precision production, harvest, and postharvest management. A prerequisite is to detect and classify fruit of different maturities from the background environment. Recently, deep learning technology has become a widely used method for intelligent fruit detection, due to it having higher accuracy, reliability, and a faster processing speed compared with traditional image-processing methods. At the same time, spectral imaging approaches can predict the maturity stage by acquiring and analyzing the spectral data of fruit samples. These maturity detection methods pay more attention to the species, such as apple, cherry, strawberry, and mango, achieving the mean average precision value of 98.7% in apple fruit. This review provides an overview of the most recent methodologies developed for in-field fruit maturity estimation. The basic principle and representative research output associated with the advantages and disadvantages of these techniques were systematically investigated and analyzed. Challenges, such as environmental factors (illumination condition, occlusion, overlap, etc.), shortage of fruit datasets, calculation, and hardware costs, were discussed. The future research directions in terms of applications and techniques are summarized and demonstrated.