An Optimization Method of Deep Transfer Learning for Vegetation Segmentation under Rainy and Dry Season Differences in a Dry Thermal Valley.

Deep learning networks might require re-training for different datasets, consuming significant manual labeling and training time. Transfer learning uses little new data and training time to enable pre-trained network segmentation in relevant scenarios (e.g., different vegetation images in rainy and dry seasons); however, existing transfer learning methods lack systematicity and controllability. So, an MTPI method (Maximum Transfer Potential Index method) was proposed to find the optimal conditions in data and feature quantity for transfer learning (MTPI conditions) in this study. The four pre-trained deep networks (Seg-Net (Semantic Segmentation Networks), FCN (Fully Convolutional Networks), Mobile net v2, and Res-Net 50 (Residual Network)) using the rainy season dataset showed that Res-Net 50 had the best accuracy with 93.58% and an WIoU (weight Intersection over Union) of 88.14%, most worthy to transfer training in vegetation segmentation. By obtaining each layer's TPI performance (Transfer Potential Index) of the pre-trained Res-Net 50, the MTPI method results show that the 1000-TDS and 37-TP were estimated as the best training speed with the smallest dataset and a small error risk. The MTPI transfer learning results show 91.56% accuracy and 84.86% WIoU with 90% new dataset reduction and 90% iteration reduction, which is informative for deep networks in segmentation tasks between complex vegetation scenes.

Citrus Huanglongbing detection and semi-quantification of the carbohydrate concentration based on micro-FTIR spectroscopy.

Citrus Huanglongbing (HLB) is nowadays one of the most fatal citrus diseases worldwide. Once the citrus tree is infected by the HLB disease, the biochemistry of the phloem region in midribs would change. In order to investigate the carbohydrate changes in phloem region of citrus midrib, the semi-quantification models were established to predict the carbohydrate concentration in it based on Fourier transform infrared microscopy (micro-FTIR) spectroscopy coupled with chemometrics. Healthy, asymptomatic-HLB, symptomatic-HLB, and nutrient-deficient citrus midribs were collected in this study. The results showed that the intensity of the characteristic peak varied with the carbohydrate (starch and soluble sugar) concentration in citrus midrib, especially at the fingerprint regions of 1175-900 cm-1, 1500-1175 cm-1, and 1800-1500 cm-1. Furthermore, semi-quantitative prediction models of starch and soluble sugar were established using the full micro-FTIR spectra and selected characteristic wavebands. The least squares support vector machine regression (LS-SVR) model combined with the random frog (RF) algorithm achieved the best prediction result with the determination coefficient of prediction ([Formula: see text]) of 0.85, the root mean square error of prediction (RMSEP) of 0.36%, residual predictive deviation (RPD) of 2.54, and [Formula: see text] of 0.87, RMSEP of 0.37%, RPD of 2.76, for starch and soluble sugar concentration prediction, respectively. In addition, multi-layer perceptron (MLP) classification models were established to identify HLB disease, achieving the overall classification accuracy of 94% and 87%, based on the full-range spectra and the optimal wavenumbers selected by the random frog (RF) algorithm, respectively. The results demonstrated that micro-FTIR spectroscopy can be a valuable tool for the prediction of carbohydrate concentration in citrus midribs and the detection of HLB disease, which would provide useful guidelines to detect citrus HLB disease.

Lychee Fruit Detection Based on Monocular Machine Vision in Orchard Environment.

Due to the change of illumination environment and overlapping conditions caused by the neighboring fruits and other background objects, the simple application of the traditional machine vision method limits the detection accuracy of lychee fruits in natural orchard environments. Therefore, this research presented a detection method based on monocular machine vision to detect lychee fruits growing in overlapped conditions. Specifically, a combination of contrast limited adaptive histogram equalization (CLAHE), red/blue chromatic mapping, Otsu thresholding and morphology operations were adopted to segment the foreground regions of the lychees. A stepwise method was proposed for extracting individual lychee fruit from the lychee foreground region. The first step in this process was based on the relative position relation of the Hough circle and an equivalent area circle (equal to the area of the potential lychee foreground region) and was designed to distinguish lychee fruits growing in isolated or overlapped states. Then, a process based on the three-point definite circle theorem was performed to extract individual lychee fruits from the foreground regions of overlapped lychee fruit clusters. Finally, to enhance the robustness of the detection method, a local binary pattern support vector machine (LBP-SVM) was adopted to filter out the false positive detections generated by background chaff interferences. The performance of the presented method was evaluated using 485 images captured in a natural lychee orchard in Conghua (Area), Guangzhou. The detection results showed that the recall rate was 86.66%, the precision rate was greater than 87% and the F1-score was 87.07%.

Citrus Tree Segmentation from UAV Images Based on Monocular Machine Vision in a Natural Orchard Environment.

The segmentation of citrus trees in a natural orchard environment is a key technology for achieving the fully autonomous operation of agricultural unmanned aerial vehicles (UAVs). Therefore, a tree segmentation method based on monocular machine vision technology and a support vector machine (SVM) algorithm are proposed in this paper to segment citrus trees precisely under different brightness and weed coverage conditions. To reduce the sensitivity to environmental brightness, a selective illumination histogram equalization method was developed to compensate for the illumination, thereby improving the brightness contrast for the foreground without changing its hue and saturation. To accurately differentiate fruit trees from different weed coverage backgrounds, a chromatic aberration segmentation algorithm and the Otsu threshold method were combined to extract potential fruit tree regions. Then, 14 color features, five statistical texture features, and local binary pattern features of those regions were calculated to establish an SVM segmentation model. The proposed method was verified on a dataset with different brightness and weed coverage conditions, and the results show that the citrus tree segmentation accuracy reached 85.27% ± 9.43%; thus, the proposed method achieved better performance than two similar methods.