Moving object segmentation (MOS) is one of the essential functions of the vision system of all robots,including medical robots. Deep learning-based MOS methods, especially deep end-to-end MOS methods, are actively inv...Moving object segmentation (MOS) is one of the essential functions of the vision system of all robots,including medical robots. Deep learning-based MOS methods, especially deep end-to-end MOS methods, are actively investigated in this field. Foreground segmentation networks (FgSegNets) are representative deep end-to-endMOS methods proposed recently. This study explores a new mechanism to improve the spatial feature learningcapability of FgSegNets with relatively few brought parameters. Specifically, we propose an enhanced attention(EA) module, a parallel connection of an attention module and a lightweight enhancement module, with sequentialattention and residual attention as special cases. We also propose integrating EA with FgSegNet_v2 by taking thelightweight convolutional block attention module as the attention module and plugging EA module after the twoMaxpooling layers of the encoder. The derived new model is named FgSegNet_v2 EA. The ablation study verifiesthe effectiveness of the proposed EA module and integration strategy. The results on the CDnet2014 dataset,which depicts human activities and vehicles captured in different scenes, show that FgSegNet_v2 EA outperformsFgSegNet_v2 by 0.08% and 14.5% under the settings of scene dependent evaluation and scene independent evaluation, respectively, which indicates the positive effect of EA on improving spatial feature learning capability ofFgSegNet_v2.展开更多
In order to obtain nozzle droplet deposition characteristics for sprayer mechanical design and variable spraying control algorithms,a nozzle droplet deposition characteristics test system for air-assisted spraying was...In order to obtain nozzle droplet deposition characteristics for sprayer mechanical design and variable spraying control algorithms,a nozzle droplet deposition characteristics test system for air-assisted spraying was designed.The test system can supply a stable wind site with precisely controlled air speed whose speed control ranges from 2 m/s to 16 m/s with maximum relative error of 4.5%.It can spray out a certain amount of liquid pesticide with adjustable spraying pressure which can be controlled with high precision while the maximum relative error is only 1.33%.The distribution of droplet deposition can be collected and measured by using the acquisition device and a pesticide deposition optical measurement system.The experiment of two-dimensional nozzle flow measurement was carried out.The results show that nozzle flow distribution is uniform and symmetric with“double-hump”shape in the spray range.Multi-nozzle overlapped droplet deposition ranges from 85%to 116%relative to the average.The nozzle droplet deposition experiment was completed.The experiment results show that in air-assisted spraying,the higher the wind speed,the less droplet deposition is affected by gravity.When the wind speed is higher than 12 m/s and spraying distance is 0.80 m,droplet deposition is concentrated on the originally designated point and hardly affected by gravity.The horizontal spray width becomes smaller with higher wind speed.When the wind speed is high,it can be considered that nozzle deposition only focuses on the nozzle center,if the position requirement is not very high in orchard spraying.展开更多
基金the National Natural Science Foundation of China(No.61702323)。
文摘Moving object segmentation (MOS) is one of the essential functions of the vision system of all robots,including medical robots. Deep learning-based MOS methods, especially deep end-to-end MOS methods, are actively investigated in this field. Foreground segmentation networks (FgSegNets) are representative deep end-to-endMOS methods proposed recently. This study explores a new mechanism to improve the spatial feature learningcapability of FgSegNets with relatively few brought parameters. Specifically, we propose an enhanced attention(EA) module, a parallel connection of an attention module and a lightweight enhancement module, with sequentialattention and residual attention as special cases. We also propose integrating EA with FgSegNet_v2 by taking thelightweight convolutional block attention module as the attention module and plugging EA module after the twoMaxpooling layers of the encoder. The derived new model is named FgSegNet_v2 EA. The ablation study verifiesthe effectiveness of the proposed EA module and integration strategy. The results on the CDnet2014 dataset,which depicts human activities and vehicles captured in different scenes, show that FgSegNet_v2 EA outperformsFgSegNet_v2 by 0.08% and 14.5% under the settings of scene dependent evaluation and scene independent evaluation, respectively, which indicates the positive effect of EA on improving spatial feature learning capability ofFgSegNet_v2.
基金China National 863 Project(2012AA101904)project 31201128 supported by NSFC,project KFZN2012W13-013IEA and project 2452013QN070 supported by Northwest A&F University.
文摘In order to obtain nozzle droplet deposition characteristics for sprayer mechanical design and variable spraying control algorithms,a nozzle droplet deposition characteristics test system for air-assisted spraying was designed.The test system can supply a stable wind site with precisely controlled air speed whose speed control ranges from 2 m/s to 16 m/s with maximum relative error of 4.5%.It can spray out a certain amount of liquid pesticide with adjustable spraying pressure which can be controlled with high precision while the maximum relative error is only 1.33%.The distribution of droplet deposition can be collected and measured by using the acquisition device and a pesticide deposition optical measurement system.The experiment of two-dimensional nozzle flow measurement was carried out.The results show that nozzle flow distribution is uniform and symmetric with“double-hump”shape in the spray range.Multi-nozzle overlapped droplet deposition ranges from 85%to 116%relative to the average.The nozzle droplet deposition experiment was completed.The experiment results show that in air-assisted spraying,the higher the wind speed,the less droplet deposition is affected by gravity.When the wind speed is higher than 12 m/s and spraying distance is 0.80 m,droplet deposition is concentrated on the originally designated point and hardly affected by gravity.The horizontal spray width becomes smaller with higher wind speed.When the wind speed is high,it can be considered that nozzle deposition only focuses on the nozzle center,if the position requirement is not very high in orchard spraying.
