Search

In this paper, a method for predicting residual film content in the cotton field plough layer based on UAV imaging and deep learning was proposed to solve the issues of high labour intensity, low efficiency, and high cost of traditional methods for residual film content monitoring. Images of residual film on soil surface in the cotton field were collected by UAV, and residual film content in the plough layer was obtained by manual sampling. Based on the three deep learning frameworks of LinkNet, FCN, and DeepLabv3, a model for segmenting residual film from the cotton field image was built. After comparing the segmentation results, DeepLabv3 was determined to be the best model for segmenting residual film, and then the area of residual film was obtained. In addition, a linear regression prediction model between the residual film coverage area on the cotton field surface and the residual film content in the plough layer was built. The results showed that the correlation coefficient (R2), root mean square error, and average relative error of the prediction of residual film content in the plough layer were 0.83, 0.48, and 11.06%, respectively. It indicates that a quick and accurate prediction of residual film content in the cotton field plough layer can be realized based on UAV imaging and deep learning. This study provides certain technical support for monitoring and evaluating residual film pollution in the cotton field plough layer.

To accurately evaluate residual plastic film pollution in pre-sowing cotton fields, a method based on modified U-Net model was proposed in this research. Images of pre-sowing cotton fields were collected using UAV imaging from different heights under different weather conditions. Residual films were manually labelled, and the degree of residual film pollution was defined based on the residual film coverage rate. The modified U-Net model for evaluating residual film pollution was built by simplifying the U-Net model framework and introducing the inception module, and the evaluation results were compared to those of the U-Net, SegNet, and FCN models. The segmentation results showed that the modified U-Net model had the best performance, with a mean intersection over union (MIOU) of 87.53%. The segmentation results on images of cloudy days were better than those on images of sunny days, with accuracy gradually decreasing with increasing image-acquiring height. The evaluation results of residual film pollution showed that the modified U-Net model outperformed the other models. The coefficient of determination(R2), root mean square error (RMSE), mean relative error (MRE) and average evaluation time per image of the modified U-Net model on the CPU were 0.9849, 0.0563, 5.33% and 4.85 s, respectively. The results indicate that UAV imaging combined with the modified U-Net model can accurately evaluate residual film pollution. This study provides technical support for the rapid and accurate evaluation of residual plastic film pollution in pre-sowing cotton fields.

Insect pest is an essential factor affecting crop yield, and the effect of pest control depends on the timeliness and accuracy of pest forecasting. The traditional method forecasts pest outbreaks by manually observing (capturing), identifying, and counting insects, which is very time-consuming and laborious. Therefore, developing a method that can more timely and accurately identify insects and obtain insect information. This study designed an image acquisition device that can quickly collect real-time photos of phototactic insects. A pest identification model was established based on a deep learning algorithm. In addition, a model update strategy and a pest outbreak warning method based on the identification results were proposed. Insect images were processed to establish the identification model by removing the background; a laboratory image collection test verified the feasibility. The results showed that the proportion of images with the background completely removed was 90.2%. Dataset 1 was obtained using reared target insects, and the identification accuracy of the ResNet V2 model on the test set was 96%. Furthermore, Dataset 2 was obtained in the cotton field using a designed field device. In exploring the model update strategy, firstly, the T_ResNet V2 model was trained with Dataset 2 using transfer learning based on the ResNet V2 model; its identification accuracy on the test set was 84.6%. Secondly, after reasonably mixing the indoor and field datasets, the SM_ResNet V2 model had an identification accuracy of 85.7%. The cotton pest image acquisition, transmission, and automatic identification system provide a good tool for accurately forecasting pest outbreaks in cotton fields.

The lateral organ boundaries domain (LBD) genes, as the plant-specific transcription factor family, play a crucial role in controlling plant architecture and stress tolerance. Although it has been thoroughly characterized in many species, the LBD family was not well studied in wheat. Here, the wheat LBD family was systematically investigated through an in silico genome-wide search method. A total of 90 wheat LBD genes (TaLBDs) were identified, which were classified into class I containing seven subfamilies, and class II containing two subfamilies. Exon–intron structure, conserved protein motif, and cis-regulatory elements analysis showed that the members in the same subfamily shared similar gene structure organizations, supporting the classification. Furthermore, the expression patterns of these TaLBDs in different types of tissues and under diverse stresses were identified through public RNA-seq data analysis, and the regulation networks of TaLBDs involved were predicted. Finally, the expression levels of 12 TaLBDs were validated by quantitative PCR (qPCR) analysis and the homoeologous genes showed differential expression. Additionally, the genetic diversity of TaLBDs in the landrace population showed slightly higher than that of the genetically improved germplasm population while obvious asymmetry at the subgenome level. This study not only provided the potential targets for further functional analysis but also contributed to better understand the roles of LBD genes in regulating development and stress tolerance in wheat and beyond.