We present a novel framework,CLIPSP,and a novel adaptive prompt method to leverage pre-trained knowledge from CLIP for scene parsing.Our approach addresses the limitations of DenseCLIP,which demonstrates the superior ...We present a novel framework,CLIPSP,and a novel adaptive prompt method to leverage pre-trained knowledge from CLIP for scene parsing.Our approach addresses the limitations of DenseCLIP,which demonstrates the superior image segmentation provided by CLIP pre-trained models over ImageNet pre-trained models,but struggles with rough pixel-text score maps for complex scene parsing.We argue that,as they contain all textual information in a dataset,the pixel-text score maps,i.e.,dense prompts,are inevitably mixed with noise.To overcome this challenge,we propose a two-step method.Firstly,we extract visual and language features and perform multi-label classification to identify the most likely categories in the input images.Secondly,based on the top-k categories and confidence scores,our method generates scene tokens which can be treated as adaptive prompts for implicit modeling of scenes,and incorporates them into the visual features fed into the decoder for segmentation.Our method imposes a constraint on prompts and suppresses the probability of irrelevant categories appearing in the scene parsing results.Our method achieves competitive performance,limited by the available visual-language pre-trained models.Our CLIP-SP performs 1.14%better(in terms of mIoU)than DenseCLIP on ADE20K,using a ResNet-50 backbone.展开更多
In the semi-physical simulation of aeroengines,using the pneumatic pressure servo control technology to provide realistic pneumatic excitation to the sensors and electronic controller can improve the confidence of the...In the semi-physical simulation of aeroengines,using the pneumatic pressure servo control technology to provide realistic pneumatic excitation to the sensors and electronic controller can improve the confidence of the simulation and reduce the test cost and risk.However,the existing methods could not satisfy the precise simulation of large-amplitude and high-frequency pulsating pressure during aeroengine surge.In this paper,a pneumatic pressure control system with asymmetric groups of the High-Speed on–off Valve(HSV)is designed,and an Improved Nonlinear Model Predictive Control(INMPC)method is proposed.First,the volumetric flow characteristics of HSV are tested and analyzed with Pulse Width Modulation(PWM)signal input.Then,a simplified HSV model with the volume flow characteristic correction is developed.Based on these,an integrated model for the whole system is further established and used as the prediction model in INMPC.To improve the computational speed of the rolling optimization process,the mapping scheme from control signal to PWM duty cycle of HSVs and the objective function with exterior penalty function are designed.Finally,the random step,sinusoidal and real engine surge data are set as the reference pressure in multiple comparative experiments to verify the effectiveness of the pressure tracking system.展开更多
文摘We present a novel framework,CLIPSP,and a novel adaptive prompt method to leverage pre-trained knowledge from CLIP for scene parsing.Our approach addresses the limitations of DenseCLIP,which demonstrates the superior image segmentation provided by CLIP pre-trained models over ImageNet pre-trained models,but struggles with rough pixel-text score maps for complex scene parsing.We argue that,as they contain all textual information in a dataset,the pixel-text score maps,i.e.,dense prompts,are inevitably mixed with noise.To overcome this challenge,we propose a two-step method.Firstly,we extract visual and language features and perform multi-label classification to identify the most likely categories in the input images.Secondly,based on the top-k categories and confidence scores,our method generates scene tokens which can be treated as adaptive prompts for implicit modeling of scenes,and incorporates them into the visual features fed into the decoder for segmentation.Our method imposes a constraint on prompts and suppresses the probability of irrelevant categories appearing in the scene parsing results.Our method achieves competitive performance,limited by the available visual-language pre-trained models.Our CLIP-SP performs 1.14%better(in terms of mIoU)than DenseCLIP on ADE20K,using a ResNet-50 backbone.
基金co-supported by the National Natural Science Foundation of China(No.51976089)the Natural Science Foundation of Fujian Province of China(No.2021J05113).
文摘In the semi-physical simulation of aeroengines,using the pneumatic pressure servo control technology to provide realistic pneumatic excitation to the sensors and electronic controller can improve the confidence of the simulation and reduce the test cost and risk.However,the existing methods could not satisfy the precise simulation of large-amplitude and high-frequency pulsating pressure during aeroengine surge.In this paper,a pneumatic pressure control system with asymmetric groups of the High-Speed on–off Valve(HSV)is designed,and an Improved Nonlinear Model Predictive Control(INMPC)method is proposed.First,the volumetric flow characteristics of HSV are tested and analyzed with Pulse Width Modulation(PWM)signal input.Then,a simplified HSV model with the volume flow characteristic correction is developed.Based on these,an integrated model for the whole system is further established and used as the prediction model in INMPC.To improve the computational speed of the rolling optimization process,the mapping scheme from control signal to PWM duty cycle of HSVs and the objective function with exterior penalty function are designed.Finally,the random step,sinusoidal and real engine surge data are set as the reference pressure in multiple comparative experiments to verify the effectiveness of the pressure tracking system.