Table 7.
Detection of pests and tree diseases. Prediction and evaluation of orchard production .
Purpose (orchard task) | Resources and discussions | Performance | References |
---|---|---|---|
Detection of pests and tree diseases | |||
Infected or diseased trees detection | ▪UAV; Faster R-CNN and Mask R-CNN approaches and fusing their outputs | ▪SEN=81.67% | (Barmpoutis et al., 2019) |
Detection of the citrus bacterial canker in disease development stages on Sugar Belle leaves and immature fruit | ▪UAV; hyperspectral camera; the neural network Radial Basis Function (RBF) and the K-nearest neighbor (KNN) | ▪ACC= 94%-100% | (Abdulridha et al., 2019) |
Identification of fruit tree pests (Tessaratoma papillosa) | ▪UAV; RGB camera; Tiny-YOLOv3 | ▪mAP= 38.12%- 95.33% | (Chen, C.J. et al., 2021) |
Detection of the degree of HLB (huanglongbing) infection on large-scale orchard citrus trees | ▪UAV; multispectral camera; stacked autoencoder (SAE) neural network | ▪ACC= 99.72% | (Deng et al., 2020) |
▪UAV; multispectral camera; autoencoder | ▪ACC=97.28%, | (Lan et al., 2020) | |
Detection of diseases in vineyards | ▪UAV; multispectral camera; LeNet-5, SegNet – single or combination | ▪ACC=78.72%-95.02 | (Kerkech et al., 2020) |
▪UAV; RGB camera; LeNet-5 | ▪ACC=95.8% | (Kerkech et al., 2018) | |
▪UAV; RGB camera; CaffeNet | ▪NA | (Bouroubi et al., 2018) | |
▪UAV; multispectral camera; VddNet | ▪ACC=93.72 | (Kerkech et al., 2020) | |
Detection of the presence and behavior of the nematode pest in coffee crops | ▪UAV; RGB camera; U-Net and PSPNet | ▪F1 = 69% | (Oliveira et al., 2019) |
Detection of black rot on grape leaves | ▪UAV; RGB camera; YOLOv3 with SPP module | ▪PRE=94.05%, SEN=93.26% | (Zhu et al., 2021) |
Sick tree detection | ▪UAV; RGB camera; different CNNs: Alexnet, Squeezenet, VGG 16; Resnet 50, Densenet 121 | ▪ACC=97.6% -99.5% | (Nguyen et al., 2021) |
Bug detection (Halyomorpha Halys) in an orchard | ▪UAV; RGB camera; processing (NN) | ▪NA | (Sorbelli et al., 2022), (Ichim et al., 2022) |
Insect detection, invasive species (Anolis carolinensis) | ▪UAV, RGB camera; SSD-based model of DCNN | ▪PRE=70% | (Aota et al., 2021) |
Invasion degree of “Solanum rostratum Dunal” detection | ▪UAV; RGB camera; DeepSolanum-Net based on U-Net | ▪F1 = 89.85% | (Wang et al., 2021) |
Prediction and evaluation of orchard production | |||
▪Method for semantic segmentation and instance segmentation of bayberry fruit. | ▪Terrestrial platform; RGB camera; Multi-module convolutional neural network | ▪AP = 75.5% -91.3% | (Lei et al., 2022) |
▪Accurate monitoring of fruit quantity in apple orchards | ▪UAV inside orchard; RGB camera; YOLO v5s | ▪AP = 90.39% | (Wang S. et al., 2022) |
▪Yield estimates in apple orchards. Detecting apples on individual trees. | ▪UAV; RGB camera; R-CNN | ▪R2 = 80% - 86% | (Apolo-Apolo et al., 2020a) |
▪Detection, counting, and estimation of the size of citrus fruits on individual trees | ▪UAV; RGB camera; Faster R-CNN | ▪F1 = 89% | (Apolo-Apolo et al., 2020b) |
▪Detection and location of longan fruits | ▪UAV; RGB camera; MobileNet backbone used to improve YOLOv4 | ▪mAP = 54.22 -89.73% | (Li D. et al., 2021) |
▪Holly fruits detection and counting | ▪UAV; RGB camera; YOLOX | ▪DR >99% | (Zhang Y. et al., 2022) |
▪Canopy extraction. Detect mango and predict the number on the tree | ▪Terrestrial platform; RGB camera; Mango YOLO, Xception, Random Forest | ▪R2 = 98% | (Koirala et al., 2021) |
▪Detect apple fruit in the orchard | ▪Manual images; RGB camera; comparing RetinaNet, Libra-RCNN, Cascade-RCNN, Faster-RCNN, FSAF, HRNet, and ATSS | ▪Maximum AP = 94.6% | (Biffi et al., 2021) |
▪Longan harvesting UAVs. Branch detection and fruit branch semantic segmentation. | ▪UAV; RGB-D camera; YOLOv5s – for detection, and improved DeepLabv3+ (MobileNet v2) for semantic segmentation | ▪ACC = 85.50% – 94.52% | (Li D. et al., 2022) |
▪Grape detection, instance segmentation | ▪RGB camera; Mask R-CNN with ResNet 101 as the backbone | ▪F1 = 91% | (Santos et al., 2020) |
▪Pear (fruit) detection | ▪RGB camera; YOLO-P | F1 = 96.1% | (Sun et al., 2023) |