Subjects: Agriculture, Forestry,Livestock & Aquatic Products Science >> Other Disciplines of Agriculture, Forestry,Livestock & Aquatic Products Science submitted time 2023-02-17 Cooperative journals: 《智慧农业（中英文）》
Abstract： Objects in farmlands often have characteristic of small volume and high density with variable light and complex background, and the available object detection models could not get satisfactory recognition results. Taking bees as research objects, a method that could overcome the influence from the complex backgrounds, the difficulty in small object feature extraction was proposed, and a detection algorithm was created for small objects irrelevant to image size. Firstly, the original image was split into some smaller sub-images to increase the object scale, and the marked objects were assigned to the sub-images to produce a new dataset. Then, the model was trained again using transfer learning to get a new object detection model. A certain overlap rate was set between two adjacent sub-images in order to restore the objects. The objects from each sub-image was collected and then non-maximum suppression (NMS) was performed to delete the redundant detection boxes caused by the network, an improved NMS named intersection over small NMS (IOS-NMS) was then proposed to delete the redundant boxes caused by the overlap between adjacent sub-images. Validation tests were performed when sub-image size was set was 300�300, 500�500 and 700�700, the overlap rate was set as 0.2 and 0.05 respectively, and the results showed that when using single shot multibox detector (SSD) as the object detection model, the recall rate and precision was generally higher than that of SSD with the maximum difference 3.8% and 2.6%, respectively. In order to further verify the algorithm in small target recognition with complex background, three bee images with different scales and different scenarios were obtained from internet and test experiments were conducted using the new proposed algorithm and SSD. The results showed that the proposed algorithm could improve the performance of target detection and had strong scale adaptability and generalization. Besides, the new algorithm required multiple forward reasoning for a single image, so it was not time-efficient and was not suitable for edge calculation.
Subjects: Agriculture, Forestry,Livestock & Aquatic Products Science >> Other Disciplines of Agriculture, Forestry,Livestock & Aquatic Products Science submitted time 2023-02-17 Cooperative journals: 《智慧农业（中英文）》
Abstract： The frequent occurrence of waterlogging stress in winter wheat not only seriously affects regional food security and ecological security, but also threatens social and economic stability and sustainable development. In order to identify the waterlogging stress level of winter wheat, a waterlogging stress gradient pot experiment was set up in this research. Three factors were controlled: waterlogging stress level (control, slight waterlogging, severe waterlogging), stress duration (5 days, 10 days, 15 days) and wheat variety (YF4, JM31, JM38). Leaf and canopy hyperspectral data were measured by using ASD Field Spec3 and Gaiasky-mini2 imaging spectrometer, respectively. The data were collected from the first waterlogging day of winter wheat. The sunny and windless weather was selected and measured every 7 days until the wheat was mature. Combined with vegetation index, normalized mean distance and spectral derivative difference entropy, if winter wheat was under waterlogging stress was monitored and stress level was identified. The results showed that: 1) the spectral response characteristics of winter wheat under waterlogging stress changed significantly in RW, RE, NIR and 1650－1800 nm region, which may be due to the sensitivity of these regions to physiological parameters affecting the spectral response characteristics, such as pigment, nutrient, leaf internal structure, etc; 2) the simple ratio pigment index SRPI was the optimal vegetation index for identifying the waterlogging stress of winter wheat. The excellent performance of this vegetation index may come from its extreme sensitivity to the epoxidation state and photosynthetic efficiency of the xanthophyll cycle pigment; 3) the red light absorption valley (RW: 640－680 nm) region was the optimal region for identifying waterlogging stress level. In RW region, waterlogging stress level of winter wheat could be determined by the spectral derivative difference entropy at heading, flowering and filling stages. The greater the level of waterlogging stress, the greater the spectral derivative difference entropy. This may be due to the fact that the RW region was more sensitive to pigment content, and the spectral derivative difference entropy could reduce the effects of spectral noise and background. This study could provide a new method for monitoring waterlogging stress, and would have a good application prospect in the precise prevention and control of waterlogging stress. There are still shortcomings in this study, such as the difference between the pot experiment and the actual field environment, the lack of independent experimental verification, etc. Next research could add pot and field experiments, combine with cross-validation, to further verify the feasibility of this research method.
Hits
Hits
Downloads
Comment