In terms of highway crack detection, the combination of UAV and deep learning network has become a powerful detection means. However, in the actual detection, in order to take into account the detection efficiency, it is necessary to ensure that the detection area is large enough, which makes the crack occupy few pixels in the image, and the image background is complex. Therefore, in this paper, DJI Mavic3 is used to establish the image data set of highway pavement cracks under complex background. And, the YOLOV5 deep learning model is improved by adding swin transformer structure and BIFPN feature pyramid. The improved YOLOV5 model achieved real-time pixel-level detection with detection accuracy of 90% and detection speed of 43.5 FPS. In terms of crack detection ability, the accuracy of the improved YOLOV5 reaches 4 pixels, and cracks of 1.2 mm can be detected in the experiment. Compared with the YOLOV7 model, the detection accuracy of the improved YOLOV5 model is increased by 15.4%. Compared with the YOLOV6 model, the calculated parameters of the improved YOLOV5 model are reduced by 59.25%. The proposed model is superior to other advanced models in crack detection.

Data processing of multi-wavelength pyrometer (MWP) is a difficult problem because unknown emissivity. So far some solutions developed generally assumed particular mathematical relations for emissivity versus wavelength or emissivity versus temperature. Due to the deviation between the hypothesis and actual situation, the inversion results can be seriously affected. So directly data processing algorithm of MWP that does not need to assume the spectral emissivity model in advance is main aim of the study. Two new data processing algorithms of MWP, Gradient Projection (GP) algorithm and Internal Penalty Function (IPF) algorithm, each of which does not require to fix emissivity model in advance, are proposed. The novelty core idea is that data processing problem of MWP is transformed into constraint optimization problem, then it can be solved by GP or IPF algorithms. By comparison of simulation results for some typical spectral emissivity models, it is found that IPF algorithm is superior to GP algorithm in terms of accuracy and efficiency. Rocket nozzle temperature experiment results show that true temperature inversion results from IPF algorithm agree well with the theoretical design temperature as well. So the proposed combination IPF algorithm with MWP is expected to be a directly data processing algorithm to clear up the unknown emissivity obstacle for MWP.

The unknown emissivity of materials is a huge obstacle in multi-wavelength pyrometry (MWP). It leads to a set of ill-posed equations that cannot be directly inverted to obtain the true temperature from a set of multi-wavelength measurements. Constraint optimization algorithms such as the gradient projection (GP) and internal penalty function (IPF) algorithms provide solutions without any emissivity model assumptions but require a narrow fixed emissivity range and an appropriate initial emissivity input value, otherwise, accuracy and computational efficiency are greatly affected. Here, we propose a generalized inverse matrix-exterior penalty function (GIM-EPF) algorithm to realize an efficient and accurate inversion without limiting the emissivity range in advance. First, a set of emissivities is obtained by the generalized inverse matrix method; these emissivities are used as initial values in the exterior penalty function iteration algorithm, from which temperature and spectral emissivity are obtained. Simulation results show that the GIM-EPF algorithm provides results superior to IPF, especially in computational efficiency. The proposed GIM-EPF method is 8 times faster than the IPF method with a 0.56% relative error at the 1800 K true temperature. The GIM-EPF method also allows near real-time diagnosis of rocket exhaust flame temperature. Rocket nozzle temperature experiment results show that the temperatures derived by the GIM-EPF algorithm agree well with the theoretical design temperature and the IPF inversion temperature.

In order to realize rapid and real temperature measurement for high temperature targets by multi-wavelength pyrometer (MWP), emissivity range constraints to optimize data processing algorithm without effect from emissivity has been developed. Through exploring the relation between emissivity deviation and true temperature by fitting of large number of data from different emissivity distribution target models, the effective search range of emissivity for every time iteration is obtained, so data processing time is greatly reduced. Simulation and experimental results indicate that calculation time is less by 0.2 seconds with 25K absolute error at 1800K true temperature, and the efficiency is improved by more than 90% compared with the previous algorithm. The method has advantages of simplicity, rapidity, and suitability for in-line high temperature measurement.

This paper introduces a novel model for low-quality pedestrian trajectory prediction, the Social Non-stationary Transformers (NSTransformers), that merges the strengths of NSTransformers and Spatio-Temporal graph transformer (STAR). The model can capture social interaction cues among pedestrians and integrate features across spatial and temporal dimensions to enhance the precision and resilience of trajectory predictions. We also propose an enhanced loss function that combines diversity loss with logarithmic root mean squared error (log-RMSE) to guarantee the reasonableness and diversity of the generated trajectories. This design adapts well to complex pedestrian interaction scenarios, thereby improving the reliability and accuracy of trajectory prediction. Furthermore, we integrate a Generative Adversarial Network (GAN) to model the randomness inherent in pedestrian trajectories. Compared to the conventional standard Gaussian distribution, our GAN approach better simulates the intricate distribution found in pedestrian trajectories, enhancing the trajectory prediction’s diversity and robustness. Experimental results reveal that our model outperforms several state-of-the-art methods. This research opens the avenue for future exploration in low-quality pedestrian trajectory prediction.