Polarization converters based on metasurfaces are one of the recently developed metadevices that can change the polarization state with designated modes, utilizing the sub-wavelength unit construction. In this paper, a kind of planar zigzag asymmetric split ring resonator (Z-ASRR) metasurface with dual bands is proposed to achieve nearly perfect polarization conversion for circularly polarized waves. Compared with the original prototype asymmetric resonant ring (ASRR), both magnitude and bandwidth have been remarkably improved for achieving a higher resonance, with the introduction of zigzag metallic wires. The reflection polarization conversion ratio possesses two peak values with 0.94 and 0.99 at 5.39 GHz and 9.65 GHz, respectively. It is also demonstrated that the introduction of extra gaps, which are closely linked with the multi-node standing wave characteristic, can control the number of resonant modes or modulate the relative bandwidth. Besides, an equivalent circuit model, the degree of zigzag bending, and the oblique incidence are further analyzed in detail. The experimental results agree well with the simulations, and this chiral metadevice could be applied for on-chip integration in an optical detection/laser, a chiral biosensor, and molecular spectroscopy.
Nowadays, object recognition and detection are inseparable from our lives which are related to various aspects in life. However, with the development of society and technology, the model of object detection become more and more complicated which affects the running velocity to some extent. In order to solve this problem, the key is to compress the model. On that note, the most important thing is to improve the running velocity without influencing the success rate of detection. To prune the mode, quantization is a low-cost way. Firstly, the object will be identified and detected by original model. After condensing model by pruning technology, run on this model to the same objects. Finally, it turns out that the model runs faster after pruning with little effect on the accuracy of object’s detection. Pruning minimizes model complexity, reduces the space needed for model storage, and also focuses on accelerating model training and prediction. It Can be applied during or after training.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.