With the increasing utilization of intelligent mobile devices for online inspection of electrical equipment in smart grids, the limited computing power and storage capacity of these devices pose challenges for deploying large algorithm models, and it is also difficult to obtain a substantial number of images of electrical equipment in public. In this paper, we propose a novel distillation method that compresses the knowledge of teacher networks into a compact few-shot classification network, employing a global and local knowledge distillation strategy. Central to our method is exploiting the global and local relationships between the features exacted by the backbone of the teacher network and the student network. We compared our method with recent state-of-the-art (SOTA) methods on three public datasets, and we achieved superior performance. Additionally, we contribute a new dataset, namely, EEI-100, which is specifically designed for electrical equipment image classification. We validated our method on this dataset and demonstrated its exceptional prediction accuracy of 94.12% when utilizing only 5-shot images.

In recent computer vision research, the pursuit of improved classification performance often leads to the adoption of complex, large-scale models. However, the actual deployment of such extensive models poses significant challenges in environments constrained by limited computing power and storage capacity. Consequently, this study is dedicated to addressing these challenges by focusing on innovative methods that enhance the classification performance of lightweight models. We propose a novel method to compress the knowledge learned by a large model into a lightweight one so that the latter can also achieve good performance in few-shot classification tasks. Specifically, we propose a dual-faceted knowledge distillation strategy that combines output-based and intermediate feature-based methods. The output-based method concentrates on distilling knowledge related to base class labels, while the intermediate feature-based approach, augmented by feature error distribution calibration, tackles the potential non-Gaussian nature of feature deviations, thereby boosting the effectiveness of knowledge transfer. Experiments conducted on MiniImageNet, CIFAR-FS, and CUB datasets demonstrate the superior performance of our method over state-of-the-art lightweight models, particularly in five-way one-shot and five-way five-shot tasks.

Adjacent segment degeneration remains a major biomechanical concern after lumbar fusion, whereas fully dynamic topping-off constructs may provide an alternative strategy by preserving segmental motion and unloading degenerated discs. In this study, a three-dimensional nonlinear finite element model of the L1-L5 lumbar spine with L3-L5 double-segment degeneration was developed to compare transforaminal lumbar interbody fusion (TLIF)-based pedicle screw fixation systems (PSFS) and Bioflex-based pedicle screw dynamic stabilization systems (PSDSS). Three interspinous process spacers, namely DIAM, Wallis, and Coflex-F, were implanted at L3-L4, and three pedicle screw diameters of 6.5, 5.5, and 4.5 mm were evaluated under flexion and extension to quantify screw-rod parameter sensitivity. The results showed that both TLIF- and Bioflex-based topping-off constructs reduced intradiscal pressure (IDP) and restricted excessive range of motion (ROM) at the transition segment, especially during extension, with a maximum L3-L4 IDP reduction of 39.49% compared with the degenerated model. Compared with fusion-based constructs, Bioflex-based PSDSS provided greater surgical-segment unloading, reducing L4-L5 IDP by 55.07% in extension and 25.30% in flexion. However, this motion-preserving effect was accompanied by higher pedicle screw stress sensitivity; in the 4.5 mm Bioflex model, the average L4 screw stress reached 15.62 MPa in flexion, representing a 51.71% increase compared with the 6.5 mm screw. In contrast, PSFS constructs showed greater stress variation in the rigid connecting rods. Overall, under the present modeling assumptions, Bioflex-based fully dynamic topping-off constructs showed more favorable disc unloading and transition-segment motion regulation than fusion-based configurations, but their biomechanical benefit should be balanced against diameter-dependent pedicle screw stability.

With the increasing utilization of intelligent mobile devices for online inspection of electrical equipment in smart grids, the limited computing power and storage capacity of these devices pose challenges for deploying large algorithm models and it’s hard to obtain a substantial number of images of electrical equipment in public. In this paper, we propose a novel distillation method that compresses the knowledge of teacher networks into a compact few-shot classification network, employing a global and local knowledge distillation strategy. Central to our method is exploiting the global and local relationship between the features exacted by the backbone of the teacher network and student network. We compare our method with recent state-of-the-art (SOTA) methods on three public datasets and achieve superior performance. Additionally, we contribute a new dataset, namely EEI-100, which is specifically designed for classification of electrical equipment. We validate our method on this dataset and demonstrate its exceptional prediction accuracy of 94.12% when utilizing only 5-shot images.