Search

At present, the cleaning of water plants in crab pond is primarily relied on manual operations, which is characterized by high labor intensity and low efficiency. The automatic water plant cleaning boat can greatly reduce the labor intensity and improve the operation efficiency, and determining the water plant distribution in crab pond is the key basis for planning the efficient operation path of the cleaning boat. Addressing the challenges posed by the high similarity between water plants and shoreline vegetation, as observed in drone-captured images of the crab pond, which complicates the accurate distinction between them, a region segmentation and localization method for water plants in crab pond based on improved YOLOv8n-seg was proposed in this paper. The focus of the improved model lies in the reduction of the model size and the enhancement of the recall rate. Firstly, in order to diminish the model size, the lightweight HGNetv2 (hierarchical graph network) based on RTDETR (real-time detection transformer) was used as the backbone feature extraction network. Secondly, the neck network was reconstructed based on the lightweight structure of Efficient Rep to reduce the number of parameters and enhance the multiscale feature fusion ability of the mode. Finally, the SegNext attention mechanism was introduced in the feature extraction layer to enhance the sensitivity of the model to the water plant area. In order to eliminate the redundant regions generated by the model in the recognition process and further improve the segmentation accuracy, the binarization processing was used to optimize the segmentation results, and the image processing algorithm was combined to screen the area of the water plant area. After coordinate conversion, the precise longitude and latitude coordinates of water plant contour were obtained. The experimental results showed that the improved model exhibited a strong discriminatory and segmentation effect on water plants in crab ponds. The parameter number, calculation quantity and model size of the improved model were only 1.49 M, 8.4 GFLOPs, and 3.27 MB, respectively. Compared with the original YOLOv8n-seg model, the parameter number, calculation quantity and model size of the improved model were reduced by 54.3%, 30.6% and 49.7%, respectively. The model before and after improvement was deployed to the Jetson Orin Nano embedded AI computer for testing. It was found that the preprocessing speed of the improved model was almost the same as that of the YOLOv8n-seg model. The inference speed, post processing speed and segment speed were increased by 30.6%, 46.4% and 32.3%, respectively. In addition, the recall, precision, and mean average precision of the improved model achieved values of 91.5%, 89.3% and 95.6%, respectively. The improved model achieved the best balance of calculation, parameters, recall and detection accuracy comparing with YOLOv5sseg, YOLOv8s-SwinTransformer, YOLOv8s-seg and YOLOv8n-GoldYOLO models. The coordinate conversion test showed that the minimum distance error of water plant location accuracy was 0.13m, the maximum distance error was 0.33 m, and the average distance error was 0.22 m, which verified that the improved model can meet the requirements of water plant region segmentation and localization in crab pond. The finding can provide an important reference for the automatic operation path planning of water plant cleaning boat. [ABSTRACT FROM AUTHOR]

Sericulture has a pivotal position in the agricultural production of China. The number of silkworms is one of the key indicators in the process of silkworm breeding, which is crucial to the feeding and healthy growth of silkworms. Accurate detection of the number of silkworms can ensure the quality of sericulture products. However, manual counting has not fully met the needs of large-scale production in recent years. There is a high demand to combine AI technology for silkworm counting. In this study, silkworm detection and counting were studied from the perspective of deep learning. Several challenges were solved in the detection of silkworms. Since the silkworms belonged to the small targets, it was difficult to detect the fewer features; When the number of silkworms was high, serious masking occurred to cause interference in the detection of densely distributed silkworms; It was difficult to ensure that the clear images of silkworms were captured from the actual production. The blur images were easy to deteriorate the accuracy of detection. Two tasks were covered as follows. 1) The image dataset of silkworms was constructed; 2) MAM-YOLO, a silkworm detection and counting model was proposed using multi-dimensional perception. The pictures of silkworms were first taken from the Guangxi Sericulture Technology Promotion Station. These pictures of silkworms were preliminarily processed. The polygon annotation was then selected, according to the body and distribution of silkworms. The pictures of silkworms were annotated with the open-source tool (Labelme). Finally, the dataset was expanded using image transformation and noise addition. The silkworm picture dataset was obtained for the subsequent training, verification, and testing. An improved YOLOv8 (MAM-YOLO) was proposed to detect the silkworm, in order to realize the accurate detection and counting. A silkworm detection and counting model was proposed using space-to-depth transformation and image deblurring. The fine-grained detection performance of blurred images was improved in the MAMYOLO model. The space-to-depth transformed convolution was utilized instead of stride convolution for the pooling layer in the feature extraction. The fine-grained features were extracted from the images, in order to further improve the detection of the low-resolution and small targets. In addition, the impact of blurred images on the detection performance was reduced by introducing a deblurring module into the model. RCS-OSA module was added to the YOLOv8 model. The network feature extraction was improved to integrate the information of different sizes in the receptive fields. The detection head of the original network was improved into the multi-dimensional sensing head. The three-dimensional features of size, space, and channel were aggregated to enhance the detection of silkworm targets. The loss function was replaced with the EIOU loss function. The regression accuracy of the detection frame was improved to reduce the object occlusion. Experimental results show that the MAM-YOLO more accurately detected the silkworm targets. The detection speed of mAP@0.5:0.95 reached 74.8% and 22.2 frames per second, indicating the better functions of silkworm detection and counting with high precision and high rate. The detection requirements were fully met in the actual work of the Guangxi sericulture station. The results show that the silkworm detection and counting system of the technical indicators met the application requirements. The finding can provide the detection services for the actual production work of the silkworm breeding station. In the end, a generalization experiment was conducted on the Global Wheat Challenge 2021 dataset. It was found that the MAM-YOLO was also achieved in the better detection performance of wheat ears, particularly on the small targets similar to dense occlusion.Sericulture has a pivotal position in China's agricultural production. The number of silkworms is a key indicator in the process of silkworm breeding, which is crucial to the feeding and healthy growth of silkworms, and accurate detection of the number of silkworms can ensure the quality of sericulture products. The traditional way of counting silkworms manually is inefficient, and it needs to be combined with technology to intelligently complete this kind of heavy work. In this thesis, from the perspective of deep learning, a silkworm detection and counting method is studied, and several problems in silkworm detection are solved: silkworms belong to small targets, which have fewer features so that they are not easy to detect; silkworms are densely distributed, and when the number of silkworms is high, serious masking phenomenon occurs which causes interference in the detection; it is difficult to ensure that the images of silkworms obtained in the actual production are clear, and blur images will make the detection affected. The main research content of this thesis is as follows: This chapter completed two main tasks: 1) Completed the production of silkworm image data set; 2) MAM-YOLO, a silkworm detection and counting model based on multi-dimensional perception, was proposed.In this paper, pictures of silkworms were taken by workers from Guangxi Sericulture Technology Promotion Station. First, these pictures of silkworms were preliminically processed. Then, according to the characteristics of the body and distribution of silkworms, the polygon annotation method was selected, and the pictures of silkworms were annotated with the open-source tool Labelme. Finally, the data set is expanded by means of image transformation and noise addition, and finally the silkworm picture data set is obtained for subsequent training, verification and testing.This chapter proposes an improved silkworm detection method based on YOLOv8, MAM-YOLO, which realizes accurate detection and counting of silkworm. Secondly, aiming at the problems of loss of fine-grained information and poor detection performance of blurred images in the MAM-YOLO model, a silkworm detection and counting model based on spaceto-depth transformation and image deblurring is proposed. The model utilizes space-to-depth transformed convolution instead of stride convolution and pooling layer in the feature extraction process, so that fine-grained features in the image can be extracted, which further improves the detection capability for low-resolution and small targets. In addition, the impact of blurred images on the detection performance is reduced by introducing a deblurring module into the model. By adding RCSOSA module to YOLOv8 model, the capability of network feature extraction is improved, and the information of different size receptive fields is integrated. The detection head of the original network is improved into multi-dimensional sensing head, and the three-dimensional characteristics of size, space and channel are aggregated to enhance the detection ability of silkworm targets. By replacing the loss function with EIOU loss function, the regression accuracy of detection frame is improved and the problem caused by object occlusion is reduced. Experiments show that MAM-YOLO can more accurately detect silkworm targets, and the detection speed of mAP@0.5:0.95 reaches 74.8% and 22.2 frames per second, which realizes the functions of silkworm detection and counting with high precision and high rate, and can meet the detection requirements in the actual work of Guangxi sericulture station. results show that the silkworm detection and counting system of the technical indicators have met the application requirements, and can provide detection services for the actual production work of silkworm breeding station.In the end, a generalization experiment was conducted on the Global Wheat Challenge 2021 data set, and it was found that MAM-YOLO also showed a high improvement in the detection performance of wheat ears, proving that the proposed method can generalize to other small targets similar to dense occlusion. [ABSTRACT FROM AUTHOR]

Autonomous navigation is an important guarantee for intelligent agricultural machinery to complete harvesting operations. Navigation line recognition is an important step in achieving autonomous navigation of intelligent agricultural machinery. Real-time and accurate detection of harvesting navigation lines can effectively accelerate the progress of operations and reduce crop loss rates. Visual sensors have the advantages of low cost and rich information acquisition, making them widely used in route detection. Frequent changes in lighting under cloudy weather make it difficult for a single feature to cope with wheat field environments. In this study, a wheat harvesting boundary detection method based on multi texture feature fusion was proposed. In the harvested area, the characteristics of low stubble and no top wheat ears result in differences in uniformity, density, and other aspects when harvesting under both light and backlight compared to the non-harvested area. At the same time, due to the exposed wheat stubble in the harvested area, it appears more regular on the image, and its gradient direction is more consistent than that in the non-harvested area. Therefore, a two-dimensional feature vector composed of image entropy and directional gradient was constructed to classify the harvested and unharvested areas of the wheat field, and then the harvesting boundary was extracted. Due to the susceptibility of image entropy features to perspective phenomena, this paper compared the entropy features of different regions in the image when fixing the camera installation angle and height, and ultimately selected the middle region of the image as the region of interest. After analyzing the characteristics of image entropy feature extraction, a histogram statistical method based on sliding windows was proposed to accelerate the speed of image entropy feature extraction. By further dividing the window to be calculated into several sub windows, sliding and combining these sub windows significantly reduced the computation. The entropy feature extraction in this paper took 0.53 seconds, which is 49.52% faster than the traditional method of directly extracting entropy from the entire window. After extracting twodimensional features, based on the distribution characteristics of the feature histogram and combined with the maximum entropy threshold segmentation algorithm, the wheat field image was preliminarily classified. For entropy features, low noise points affected the threshold segmentation effect. By comparing the impact of discarding different proportions of low entropy points on the segmentation threshold, this paper discarded 10% of low entropy points, enhancing the threshold segmentation effect. The method of removing small connected regions was used to eliminate the misclassified regions in the initial classification binary image. Then, the Canny operator was used to extract edge contour points distributed near the harvesting boundary. Due to the consistent direction between the harvesting boundary and the agricultural machinery's forward direction, the Ransac algorithm was used to restrict the fitting line in the region, obtaining the accurate harvesting boundary. In order to verify the feasibility of the method proposed in this article, wheat field images under different lighting conditions were collected, including 2 200 weak light images, 758 local strong light images, and 1 134 strong light images. After selecting the region of interest, the size of the image to be processed was 240 pixels×1 280 pixels. The algorithm in this paper took an average of 0.88 s to process each image on laptops. The detection rates under weak light, local strong light, and strong light were 90.41%, 88.26%, and 89.68%, respectively, with an average of 89.45%. Compared with the traditional method of using Adaboost algorithm for machine learning, the detection speed of this algorithm was improved by 73.89%, and the detection rate was increased by 46.19 percentage points, 46.00 percentage points, and 49.64 percentage points under weak light, local strong light, and strong light, respectively. Compared with traditional algorithms, the method proposed in this paper significantly improved real-time performance and detection rate, which can provide reference for the detection of field agricultural machinery navigation lines. [ABSTRACT FROM AUTHOR]

Agricultural named entity recognition is a fundamental task for information extraction in the agricultural domain. Aiming at the problems of local context features、unable to solve the polysemy of the word、low recognition rate of rare entities in the process of entity recognition, the model combined with character level features and dictionary feature was proposed to automatically identify entities in the text，the character level features were obtained from the BERT(Bidirectional Encoder Representations from Transformers)model. Firstly, the BERT pre-trained language model was used to integrate the left and right contextual information to obtain the character level features, enhance the semantic representation of words, in order to alleviate the problem of polysemy; Secondly, we built an agricultural dictionary and introduced external dictionary information through the feature extraction strategy to improve the recognition accuracy of the model for rare or unknown entities. Among them, two feature extraction strategies were designed to capture the dictionary features, included N-gram feature template algorithm and bi-direction maximum matching algorithm. Then, the character level features and dictionary features were fused as the input of the next neural network layer. Finally, the fused feature information were encoded by the BiLSTM (Bi-directional Long-short Term Memory) neural network layer, obtained the sequence feature matrix, and the optimal text label sequence was obtained by CRF (Conditional Random Field). Based on the knowledge of domain experts, a labeling strategy of named entities in the agricultural field was proposed to solve the problem of fuzzy boundaries of agricultural named entities, in order to ensure the integrity of the entities. The experiments were carried out on the corpus of agricultural, which contained 5 295 labeled corpora and 5 categories of agricultural entities. The results showed that better overall performance was achieved in the corpus, where the recognition precision, recall, and F1-score were 94.84%, 95.23%, and 95.03%, respectively. In terms of specific categories, due to the obvious boundary characteristics of crop diseases and pesticide, the model achieved higher recognition precision than the remaining three entities of agricultural, such as machinery, pests, and crop variety. Experimental comparison showed that for the effectiveness of the dictionary feature extraction strategy, the performance of the model based on the bi-direction maximum matching algorithm was better than the N-gram feature template algorithm. When the number of templates was 10, the performance of the model based on N-gram feature template was the best with the recognition precision of93.95%and F1-score of 94.03%. The bi-directional maximum matching algorithm using feature embedding can obtain more potential information, which was better than one-hot encoding. The precision and F1-score of the model were improved by 0.49 and 0.91 percentage points, respectively. Compared with the models based on BiLSTM-CRF, BERT-BiLSTM-CRF, the precision of the BERT-Dic-BiLSTM-CRF model proposed in this paper had obvious performance advantages with the highest recognition precision of 94.84%. Compared with the BERT-BiLSTM-CRF model, for the recognition performance of rare or unknown entities, the recognition precision of the BERT-Dic-BiLSTM-CRF model was improved by 5.93 and 6.44 percentage points, respectively. Further verifying that the integration of dictionary features into the model can improve the recognition accuracy of the model for such entities. [ABSTRACT FROM AUTHOR]

High-quality optical data is difficult to acquire in southwest China, due to the cloudy and rainy weather. The synthetic aperture radar (SAR) can work for all-time and all-weather, but the revisit period is usually long. Therefore, it is a high demand for the synergy of optical and radar data rather than a single remote sensing data source, particularly for the survey and assessment on the land resource, as well as the global change. This study performed an object-oriented guided segmentation on the difference images generated from 2016 and 2019 quad-pol Radarsat-2 data using the 2019 Sentinel-2 optical data of Meishan region. A Fractal Net Evolution Approach (FNEA) was first utilized to segment the 2016 optical image for an initial segmentation. Then, the second segmentation was implemented using the same approach, where the features of the SAR image were provided for the purpose of guided segmentation. As such, the regions of growth were merged from small to large, in order to eventually form a complete geographical object. A final segmentation was obtained to select the changed and unchanged samples. More importantly, an object contained both changed and unchanged pixels, particularly with a single type of pixel inside the samples. Then, the object-level multidimensional features were extracted using the samples. The initial radar parameters included the backward scattering coefficient, Pauli and Freeman-Durden polarization decomposition parameters. The segmentation was utilized to generate the geographic objects. Meanwhile, some parameters were calculated using the initial parameters, including the mean, and standard deviation, whereas, the texture feature parameters using the Gray-level co-occurrence matrix (GLCM) included the mean, standard deviation, homogeneity, contrast, entropy, and the correlation for a total of 72-dimensional features. The redundant feature dataset was filtered to select the extracted feature set. Consequently, an FSO-RF change detection framework was proposed to achieve the feature optimization and the acquisition of the final change detection. The inter-sample distance metric learning was used to optimize the separability of feature space. The optimal features were used to calculate the distance between the changed and unchanged samples in different features. Finally, the change detection task was implemented using an object-oriented random forest classifier. The existing change detection pixel-based methods were selected to verify the improved model, including the change vector analysis (CVA), principal component analysis (PCA), multivariate alteration detection (MAD), Iteration multivariate alteration detection (IR-MAD), support vector machines (SVM), and random forest (RF) change detection using radar difference image segmentation. The change detection framework showed a remarkable improvement, with an overall accuracy of 92.90% in the test area. Two representative areas were selected in the study area for sampling and validation. The experiments demonstrated that better performance was achieved than that of the traditional, although the accuracy indexes were degraded in some areas with the complex features. The overall accuracies of the two sample areas reached 95.08 % and 88.16%. The algorithm in this paper can well meet the needs of detecting changes in different feature categories such as towns and farmlands, and has certain application value in the monitoring of national land resources. [ABSTRACT FROM AUTHOR]