Fruit Yield Prediction Research Articles

Fruit yield prediction and estimation in orchards: A state-of-the-art comprehensive review for both direct and indirect methods

• A systematic review of orchard yield prediction and estimation studies. • Summary of input features and modeling algorithms for indirect yield prediction. • Summary of applied platforms and counting strategies for direct yield estimation. • Application of engineering technologies needs coordinating with agronomic. • Recommendations of promising technologies to be use for different planting areas. Orchard pre-harvest yield data is important for fruit growers, which can be used for economic benefit evaluation, management mode adjustment and so on. However, traditional manual operation by sampling estimation is quite an onerous and time-consuming task. The main approach of automatic yield monitoring is by establishing multi-information comprehensive prediction systems or using intelligent equipment. A review is performed to investigate and analyze the past 12 years (from 2010 to 2021) of research work regarding orchard yield prediction and estimation. According to our investigation, the most widely used input features in yield prediction systems are various vegetation indices information of plants, while machine learning is the most modeling method applied. In addition, machine vision systems based on image processing and deep learning have been developed rapidly in the field of agriculture and is also widely applied in orchard yield prediction. Finally, major challenges and countermeasures for orchard yield prediction and estimation are discussed. Complicated natural environments with inconsistent horticultural management practices are still critical challenges for large scale commercialization of researched yield prediction methods. Therefore, combination of engineering technology optimization and standardization of agronomic management practices are necessary to realize automatic operation in complex agricultural fields. The review is intended to summarize the development of yield prediction and estimation technologies and provided suggestions for orchard intelligent management.

Polyphyletic Loss: Litchi Flower Detection with Occlusion

Controlling shoot and promoting flower of litchi is one of the key technologies for the high yield of fruit trees, but the occlusion caused by dense growth has a great impact on the detection of flowers. To solve the problem, this paper designs a polyphyletic loss function to detect the occlusive litchi flowers. A new aggregation loss term is proposed in the loss function to force the proposal box to approach and compactly locate the corresponding target. At the same time, the segmentation loss of the bounding box specially designed for the dense crop scene is added to keep the prediction box away from the surrounding objects and improve the robustness of detecting a large number of flowers. We conducted experiments on the self-built litchi flower data set and the multi species fruit flower data set at the same time. Compared with other methods, the proposed method has higher robustness and detection accuracy, which provides an important idea and method for flower number statistics and fruit yield prediction in dense scenes.

Modeling the final fruit yield of coriander (Coriandrum sativum L.) using multiple linear regression and artificial neural network models

ABSTRACT The prediction of fruit yield in the next generation is one of the most important breeding objectives in agricultural research. For this purpose, different generations of coriander consisted of six quietly divergent parents, their 15 F1 hybrids and 15 F2 families were evaluated during the 2014–2017 growing seasons. The artificial neural network (ANN) models were constructed to predict the fruit yield using morphological and agronomic factors, and compare the performance of ANN models with multiple linear regression (MLR) models. According to the principal component analysis (PCA) and stepwise regression (SWR), four traits of days to flowering, thousand fruit weight, fertile umbel number per plant and branch number per plant were selected as input variables in both ANN and MLR models. A network with Levenberg–Marquart learning algorithm, SigmoidAxon transfer function, one hidden layer with four neurons and having 0.461 root-mean-square error (RMSE), 0.335 mean absolute error (MAE) and 0.938 determination coefficient (R2) selected as the final ANN model. The ANN model was a more accurate tool rather than MLR for predicting fruit yield in coriander. According to sensitivity analysis, days to flowering and thousand fruit weight traits were identified as the most effective characters in fruit yield.

Robotic real-time 3D object reconstruction using multiple laser range finders

An accurate 3D model of an outdoor scene can be used in many different scenarios of precision agriculture, for instance to analyse the silhouette of a tree crown canopy for precision spraying, to count fruit for fruit yield prediction or to simply navigate a vehicle between the plant rows. Instead of using stereovision, limited by the problems of different light intensities, or by using expensive multi-channel 3D range finder (LIDAR scanner), limited by the number of channels, this work investigates the possibility of using two single channel LIDAR scanners mounted on a small robot to allow a real-time 3D object reconstruction of the robot environment. The approach used readings captured by two LIDAR scanners, SICK LMS111 and SICK TiM310, where the first one was scanning horizontally and the second one vertically. In order to correctly map the 3D points of the readings from the vertical sensor into a 3D space, a custom SLAM algorithm based on image registration techniques was used to calculate the new positions of the robot. The approach was tested in an indoor and outdoor environment, proving its accuracy with an error rate of 0.02 m±0.02 m for vertical and −0.01 m±0.13 m for the horizontal plane.

Model Selection for Nondestructive Quantification of Fruit Growth in Pepper

Quantifying fruit growth can be desirable for several purposes (e.g., prediction of fruit yield and size, or for the use in crop simulation models). The goal of this article was to determine the best sigmoid function to describe fruit growth of pepper (Capsicum annuum) from nondestructive fruit growth measurements. The Richards, Gompertz, logistic, and beta growth functions were tested. Fruit growth of sweet pepper was measured nondestructively in an experiment with three different average daily temperatures (18, 21, and 24 °C) and in an experiment with six cultivars with different fruit sizes (20 to 205 g fresh weight). Measurements of fruit length and fruit diameter or circumference were performed twice per week. From these, fruit volume was estimated. A linear relationship related fruit fresh weight to estimated fruit volume, and a Ricker or polynomial function related fruit dry matter content to fruit age. These relations were used to convert estimated fruit volume into fruit fresh and dry weights. As dry weight increased until harvest, fitting the sigmoid function to the dry weight data was less suitable: it would create uncertainty in the estimated asymptote. Therefore, the sigmoid functions were fitted to fresh weight growth of the fruit. The Richards function was the best function in each data set, closely followed by the Gompertz function. The fruit dry weight growth is obtained by multiplication of the sigmoid function and the function relating fruit dry matter content to fruit age.