Image/video processing for fruit detection in the tree using hard-coded feature extraction algorithms has shown high accuracy on fruit detection during recent years. While accurate, these approaches even with high-end hardware are still computationally intensive and too slow for real-time systems. This paper details the use of deep convolution neural networks architecture based on single-stage detectors. Using deep-learning techniques eliminates the need for hard-code specific features for specific fruit shapes, color and/or other attributes. This architecture takes the input image and divides into AxA grid, where A is a configurable hyper-parameter that defines the fineness of the grid. To each grid cell an image detection and localization algorithm is applied. Each of those cells is responsible to predict bounding boxes and confidence score for fruit (apple and pear in the case of this study) detected in that cell. We want this confidence score to be high if a fruit exists in a cell, otherwise to be zero, if no fruit is in the cell. More than 100 images of apple and pear trees were taken. Each tree image with approximately 50 fruits, that at the end resulted on more than 5000 images of apple and pear fruits each. Labeling images for training consisted on manually specifying the bounding boxes for fruits, where (x, y) are the center coordinates of the box and (w, h) are width and height. This architecture showed an accuracy of more than 90% fruit detection. Based on correlation between number of visible fruits, detected fruits on one frame and the real number of fruits on one tree, a model was created to accommodate this error rate. Processing speed is higher than 20 FPS which is fast enough for any grasping/harvesting robotic arm or other real-time applications. HIGHLIGHTS Using new convolutional deep learning techniques based on single-shot detectors to detect and count fruits (apple and pear) within the tree canopy.

This work describes a fruit gauge based on a low-cost linear potentiometer interfaced to a data-logger, which allows continuous measurement of fruit diameter. The sensor is supported by a custom-built stainless steel frame designed to be easily applied to different size fruits. Highly significant linear relationships between tension (mV) and displacement (mm) have been found during sensor calibration. The average slope of this relationship, 5.4 × 10 ⁻³ with a standard deviation of 5 × 10 ⁻⁵ , was used as conversion coefficient. Temperature stability was tested by submitting both the sensor and the whole gauge to a temperature range of 5–60 °C. The maximum deflection found was ±1 mV. Measurements performed during the three developmental stages of peach [ Prunus persica (L), Batsch.] fruit showed high sensitivity of the gauge which allowed clear detection of diurnal patterns of fruit diameter changes and precise monitoring of minute variations in fruit growth rates during 24 h.

The effect of shading (i.e. reduction of sunlight availability) on fruit growth physiology has been widely studied in apple (Malus domestica Borkh.), but little knowledge exist about fruit growth responses to changes in the light spectrum. The aim of the present research was to study the effect of use of colored nets with differential sunlight transmission in the blue (B, 400-500 nm), red (R, 600-700 nm) and far-red (FR, 700-800 nm) spectra on apple fruit growth and physiological associated responses. Three year old ’Fuji’ apple trees were covered with 40% photo-selective blue and red shade nets, 40% neutral grey shade net, and 20% neutral white net as control. Red and blue net reduced in the same proportion (27%) the photosynthetically active radiation with respect to control. However, blue net increased by 30% and reduced by 10% the B:R and R:FR the light relations, respectively. Maximal fruit growth rate under blue and grey nets was 15-20% greater than control. Fruit weight under blue net was 17% greater than control, but no significant differences in fruit weight were found among red net and control. Leaf photosynthesis and total leaf area under blue net were 28% and 30% higher than control, respectively; with ensuing positive effect on tree net C assimilation rate and total dry matter production. Results suggest that shifting the B, R, and FR light composition with photo-selective nets could be a useful tool to manipulate the photosynthetic and morphogenetic process regulating the carbohydrate availability for apple fruit growth.

Yield maps provide a detailed account of crop production and potential revenue of a farm. This level of details enables a range of possibilities from improving input management, conducting on-farm experimentation, or generating profitability map, thus creating value for farmers. While this technology is widely available for field crops such as maize, soybean and grain, few yield sensing systems exist for horticultural crops such as berries, field vegetable or orchards. Nevertheless, a wide range of techniques and technologies have been investigated as potential means of sensing crop yield for horticultural crops. This paper reviews yield monitoring approaches that can be divided into proximal, either direct or indirect, and remote measurement principles. It reviews remote sensing as a way to estimate and forecast yield prior to harvest. For each approach, basic principles are explained as well as examples of application in horticultural crops and success rate. The different approaches provide whether a deterministic (direct measurement of weight for instance) or an empirical (capacitance measurements correlated to weight for instance) result, which may impact transferability. The discussion also covers the level of precision required for different tasks and the trend and future perspectives. This review demonstrated the need for more commercial solutions to map yield of horticultural crops. It also showed that several approaches have demonstrated high success rate and that combining technologies may be the best way to provide enough accuracy and robustness for future commercial systems.

The kiwifruit berry is characterized by an early stage of rapid growth, followed by a relatively long stage of slow increase in size. Vascular and transpiration flows are the main processes through which water and carbon enter/exit the fruit, determining the daily and seasonal changes in fruit size. This work investigates the biophysical mechanisms underpinning the change in fruit growth rate during the season. The daily patterns of phloem, xylem and transpiration in/outflows have been determined at several stages of kiwifruit development, during two seasons. The different flows were quantified by comparing the diurnal patterns of diameter change of fruit, which were then girdled and subsequently detached while measurements continued. The diurnal courses of leaf and stem water potential and of fruit pressure potential were also monitored at different times during the season. Xylem and transpiration flows were high during the first period of rapid volume growth and sharply decreased with fruit development. Specific phloem import was lower and gradually decreased during the season, whereas it remained constant at whole-fruit level, in accordance with fruit dry matter gain. On a daily basis, transpiration always responded to vapour pressure deficit and contributed to the daily reduction of fruit hydrostatic pressure. Xylem flow was positively related to stem-to-fruit pressure potential gradient during the first but not the last part of the season, when xylem conductivity appeared to be reduced. The fruit growth model adopted by this species changes during the season due to anatomical modifications in the fruit features.