Search

Instance segmentation is crucial to modern agriculture and the management of pig farms. In practical farming environments, challenges arise due to the mutual adhesion, occlusion, and dynamic changes in body posture among pigs, making accurate segmentation of multiple target pigs complex. To address these challenges, we conducted experiments using video data captured from varying angles and non-fixed lenses. We selected 45 pigs aged between 20 and 105 days from eight pens as research subjects. Among these, 1917 images were meticulously labeled, with 959 images designated for the training set, 192 for validation, and 766 for testing. To enhance feature utilization and address limitations in the fusion process between bottom-up and top-down feature maps within the feature pyramid network (FPN) module of the YOLACT model, we propose a pixel self-attention (PSA) module, incorporating joint channel and spatial attention. The PSA module seamlessly integrates into multiple stages of the FPN feature extraction within the YOLACT model. We utilized ResNet50 and ResNet101 as backbone networks and compared performance metrics, including AP0.5, AP0.75, AP0.5-0.95, and AR0.5-0.95, between the YOLACT model with the PSA module and YOLACT models equipped with BAM, CBAM, and SCSE attention modules. Experimental results indicated that the PSA attention module outperforms BAM, CBAM, and SCSE, regardless of the selected backbone network. In particular, when employing ResNet101 as the backbone network, integrating the PSA module yields a 2.7% improvement over no attention, 2.3% over BAM, 2.4% over CBAM, and 2.1% over SCSE across the AP0.5-0.95 metric. We visualized prototype masks within YOLACT to elucidate the model’s mechanism. Furthermore, we visualized the PSA attention to confirm its ability to capture valuable pig-related information. Additionally, we validated the transfer performance of our model on a top-down view dataset, affirming the robustness of the YOLACT model with the PSA module.

Copyright of INMATEH - Agricultural Engineering is the property of INMATEH - Agricultural Engineering and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

This study proposed a method for detecting lameness in dairy cows based on machine vision, addressing the challenges associated with manual detection. Data from a dairy farm in Taigu, Shanxi, China were collected and divided into two parts. The first part was utilized to precisely position the cow's back by employing a dedicated deep learning model named GhostNet_YOLOv4, which can be implemented on mobile or embedded devices. The second part was used with the Visual Background Extractor (Vibe) algorithm, incorporating additional morphological processing techniques. Enhancing the Vibe algorithm, a widely used background subtraction algorithm for image sequences, achieved more accurate recognition of the specific pixel areas of cows. Subsequently, cow shape-related feature parameters were extracted from the back area using the combined approach. These parameters were used to calculate the average curvature, which describes the degree of curvature of the cow's back contour during walking. The differences in curvature values were employed for classification to detect lameness. Through extensive experimentation, distinct average curvature ranges of [-0.025, -0.125], [-0.025, +∞], and [-∞, -0.125] were established for normal cows, early lameness, and moderate-severe lameness, respectively. The algorithm's effectiveness was validated by processing 600 image sequences of dairy cows, resulting in a lameness detection accuracy of 91.67%. These findings can serve as a reference for the timely and accurate recognition of lameness in dairy cows. [ABSTRACT FROM AUTHOR]

The automatic detection of wheat ears in the field has important scientific research value in yield estimation, gene character expression and seed screening. The manual counting method of wheat ears commonly used by breeding experts has some problems, such as low efficiency and high influence of subjective factors. In order to accurately detect the number of wheat ears in the field, based on MobileNet series network model, deep separable convolution module and alpha channel technology, the YOLOv4 model is reconstructed and successfully applied to the task of wheat ear yield estimation in the field. The model can adapt to the accurate recognition and counting of wheat ear images in different light, viewing angle and growth period. At the same time, the model volume with different alpha parameters is more suitable for mobile terminal deployment. The results show that the parameters of the improved YOLOv4 model are five times smaller than the original model, the average detection accuracy is 76.45%, and the detection speed FPS is two times higher than the original model, which provides accurate technical support for rapid yield estimation of wheat in the field. [ABSTRACT FROM AUTHOR]

Apple planting process is often accompanied by the impact of a variety of diseases. A single apple leaf often presents the situation of multiple diseases occurring at the same time, which brings great challenges to fruit farmers' rapid diagnosis and correct control. In this paper, aiming at the rapid detection and recognition of multi-category apple leaf disease, a multi-target detection model is constructed to realize the rapid detection and recognition of single leaf and multi leaf, single disease and multi disease. Through the technical means of manual labeling, data enhancement and parameter optimization, Yolo v4, SSD and EfficientDet are selected to train and evaluate the apple leaf disease data set. The results show that the target detection model based on Yolo v4 achieves better training effect, and its mAP value is 83.34%. The model can meet the needs of rapid disease spot detection and recognition of single leaf single disease and multi leaf multi disease in natural environment. [ABSTRACT FROM AUTHOR]

The acquisition of traditional wheat ear phenotypic parameters is labour intensive and subjective, and some trait parameters are difficult to measure, which greatly limits the progress of wheat ear research. To obtain the phenotypic parameters of wheat ears in batches at a low cost, this paper proposed a convenient and accurate method for extracting phenotypic parameters of wheat ears. First, three improvement directions were proposed based on the Mask Region-Convolutional Neural Network (Mask-RCNN) model. 1) To extract the multiscale features of wheat ears, a hierarchical residual link was constructed in a single residual block of the backbone network ResNet101 to obtain information on different sizes of receptive fields. 2) The feature pyramid network (FPN) was improved to increase the recognition accuracy of wheat ear edges through multiple two-way information flow sampling. 3) The mask evaluation mechanism was improved, specific network blocks were used to learn and predict the quality of the mask, and the detection of wheat ears and grains was performed by precise segmentation; an automatic extraction algorithm was designed for wheat ear phenotypic parameters based on the segmentation results to extract 22 phenotypic parameters. The experiments showed that the improved Mask-RCNN was superior to the existing model in the segmentation accuracy of wheat ears and grains; the parameters of wheat ear length, width, and number of grains extracted by the automatic extraction algorithm were close to the manual measurement values. This research meets the demand for automatic extraction of wheat ear phenotype data for large-scale quality testing and commercial breeding and has strong practicability. [ABSTRACT FROM AUTHOR]