Showing papers by "Hui Li published in 2023"
02 Jan 2023
TL;DR: In this paper , a visual transformer based video super resolution (ViTSR) network was proposed to reconstruct high resolution HR videos frames, which achieved an accuracy of 96.34% on the multi-objects extraction task and can be applied to different AM processes.
Abstract: —In-situ monitoring system can be used to monitor the quality of additive manufacturing (AM) processes. In the case of digital image correlation (DIC) based in-situ monitoring systems, high-speed cameras were used to capture images of high resolutions. This paper proposed a novel in-situ monitoring system to accelerate the process of digital images using artificial intelligence (AI) edge computing board. It built a visual transformer based video super resolution (ViTSR) network to reconstruct high resolution (HR) videos frames. Fully convolutional network (FCN) was used to simultaneously extract the geometric characteristics of molten pool and plasma arc during the AM processes. Compared with 6 state-of-the-art super resolution methods, ViTSR ranks first in terms of peak signal to noise ratio (PSNR). The PSNR of ViTSR for 4× super resolution reached 38.16 dB on test data with input size of 75 pixels × 75 pixels. Inference time of ViTSR and FCN was optimized to 50.97 ms and 67.86 ms on AI edge board after operator fusion and model pruning. The total inference time of the proposed system was 118.83 ms, which meets the requirement of real-time quality monitoring with low cost in-situ monitoring equipment during AM processes. The proposed system achieved an accuracy of 96.34% on the multi-objects extraction task and can be applied to different AM processes.
TL;DR: Wang et al. as discussed by the authors proposed a scale-aware two-stage high dynamic range imaging framework (STHDR) to generate high-quality ghost-free HDR image, which can progressively and effectively improve the HDR composition performance.
Abstract: Deep high dynamic range (HDR) imaging as an image translation issue has achieved great performance without explicit optical flow alignment. However, challenges remain over content association ambiguities especially caused by saturation and large-scale movements. To address the ghosting issue and enhance the details in saturated regions, we propose a scale-aware two-stage high dynamic range imaging framework (STHDR) to generate high-quality ghost-free HDR image. The scale-aware technique and two-stage fusion strategy can progressively and effectively improve the HDR composition performance. Specifically, our framework consists of feature alignment and two-stage fusion. In feature alignment, we propose a spatial correct module (SCM) to better exploit useful information among non-aligned features to avoid ghosting and saturation. In the first stage of feature fusion, we obtain a preliminary fusion result with little ghosting. In the second stage, we conflate the results of the first stage with aligned features to further reduce residual artifacts and thus improve the overall quality. Extensive experimental results on the typical test dataset validate the effectiveness of the proposed STHDR in terms of speed and quality.
TL;DR: In this article , a mixed integer program-ming model is established to schedule all operating facilities and equipment at an auto-mated landmaritime multimodal container terminal with multi-size containers, in which operating facilities such as quay cranes, vehicles, yard cranes and external container trucks are involved.
Abstract: With the increasing volume of shipping containers, container multimodal transport and port scheduling have attracted much attention. The allocation and dispatching of handling equipment to minimize completion time and energy consumption have always been a focus of research. This paper considers a scheduling problem at an auto-mated landmaritime multimodal container terminal with multi-size containers, in which operating facilities and equipment such as quay cranes, vehicles, yard cranes, and external container trucks are involved. Moreover, the diversity of container sizes and the location of handshake areas in yards are concerned. A mixed integer program-ming model is established to schedule all operating facilities and equipment. To solve the mathematical model is a NP-hard problem, which is difficult to be solved by conventional methods. Then we propose a heuristic algorithm which merges multiple targets into one and designs an improved genetic algorithm based on the heuristic combi-nation strategy in which 20-ft containers are paired-up to the same yard before allocation. After that, some exper-iments are designed to prove the effectiveness of the model and the algorithm. The effect of configurations on efficiency and energy consumption under different conditions is discussed, and the influences of different parame-ters and the proportion of 20-ft containers are also compared. Furthermore, the influence of locations of hand-shake area with different yard quantities are compared. To conclude, there is an optimal number of equipment to be allocated. If few equipment is used, the operation time will be prolonged; if too many, the energy consumption will be increased. When the yard operation is the bottleneck, the handover location should be in the centre, other-wise other locations might be feasible. When the proportion of 20-ft containers that can be combined is large, the method proposed in this paper has advantages over traditional methods. The proposed algorithm has made a breakthrough in improving efficiency and reducing energy consumption.