Sports analytics is a field of study that utilizes camera and sensor data to monitor the athlete’s performance and health to optimize the player's strategy and increase the success rate. Coaches rely on analytics to scout opponents and optimize play calls in gameplay. With the advancement in artificial intelligence, accessible and in-depth data collection has been enabled. The well-grounded technique for performance evaluation in sports analytics is Human Pose Estimation (HPE). Our focus is on real-time action recognition in combat sports like boxing. Existing state-of-the-art deep learning models are heavily parameterized, so can’t be used in real-time in any low-end devices. Apart from this, fine-grained classification in highly dynamic activities in sports are typically performed using sensors only. Our proposed Machine Learning based pipeline provides real-time fine-grained solution for 14 boxing punch types of classification using RGB video only. Our approach includes the implementation of three novel and generalized motion dynamics features that encode velocity as well as acceleration of the pose sequences., 1) Unified-Axis Angular Encoding (UAE), 2) 2D Motion Dynamics Descriptors (2DMDD), 3) Fifth-order Angular Encoding (FAE). We employed classical machine learning algorithms I.e., Support Vector Machine (SVM), Random Forest (RF), and K Nearest Neighbours (KNN) to make a lightweight model and test it on YouTube videos. The average accuracies of pipeline using the proposed features are found to be 55%, 92% and 84% for UAE, 2DMDD, and FAE respectively. Using KNN, we have achieved 99% accuracy on 10-fold cross-validation by using FAE features.
The capabilities of modern precision nanofabrication and the wide choice of materials [plasmonic metals, high-index dielectrics, phase change materials (PCM), and 2D materials] make the inverse design of nanophotonic structures such as metasurfaces increasingly difficult. Deep learning is becoming increasingly relevant for nanophotonics inverse design. Although deep learning design methodologies are becoming increasingly sophisticated, the problem of the simultaneous inverse design of structure and material has not received much attention. In this contribution, we propose a deep learning-based inverse design methodology for simultaneous material choice and device geometry optimization. To demonstrate the utility of the proposed method, we consider the topical problem of active metasurface design using PCMs. We consider a set of four commonly used PCMs in both fully amorphous and crystalline material phases for the material choice and an arbitrarily specifiable polygonal meta-atom shape for the geometry part, which leads to a vast structure/material design space. We find that a suitably designed deep neural network can achieve good optical spectrum prediction capability in an ample design space. Furthermore, we show that this forward model has a sufficiently high predictive ability to be used in a surrogate-optimization setup resulting in the inverse design of active metasurfaces of switchable functionality.
In recent years, there has been a growing interest in active metasurfaces. In particular, phase change material-based metasurfaces offering all-optical reconfigurability are being explored. Despite recent progress, further improvement in device reconfiguration energies and optical contrast achievable between the amorphous and crystalline states is desirable. In this work, we demonstrate that using a mirror-backed chalcogenide-based narrowband perfect absorber metasurface can significantly improve the device’s reflection contrast at much lower energies than its mirrorless case. By considering a GST225 metasurface operating in the near IR, our systematic numerical study finds improved reflection contrast (up to −32 dB, Q-factor 19.22 compared with 9.59 dB, Q-factor 11 for the mirrorless case). For the mirrored case, the thermal study finds faster crystallization (up to 6 times) at reduced reconfiguration thresholds (72 times lower) compared with the mirrorless case. This results in a more than 2 orders of magnitude higher device figure of merit [defined as the change in reflection contrast (in dB) to a corresponding change in optical energy (in nJ)] compared with the mirrorless case. The results are promising for high-performance metasurfaces at reduced switching energies.
A platform of recent interest and large application potential is one in which the light emitters are directly integrated with an optical metasurface. Plasmonic metals and high-refractive-index, dielectric Mie metasurfaces have been explored in this connection but have their challenges. We propose low-refractive-index, contrast nanostructured thin films for studying metasurface-emitter systems, specifically focusing on two configurations of practical importance: LED white light generation using color converting polymers and fluorescence-based sensing in aqueous media. To achieve light emission enhancement in a low-index contrast, low optical absorption setting, we exploit the excitation of quasi-bound states in the continuum modes in a mirror-symmetry broken grating. Our numerical study predicts widely tuneable sharp-linewidth emission enhancements, near-zero quenching, and large and controllable active volume in the grating vicinity, which are significant improvements in comparison with both plasmonic and high-index contrast, all-dielectric platforms. When compared with simple gratings, mirror-symmetry broken gratings give four times larger radiative enhancement. Our results are of interest in furthering experimental activity and realizing applications such as light converter for efficient white LEDs and smart detection electronics-integrated substrates for sensing.
State-of-the-art nanofabrication permits the realization of highly aligned multi-layered metasurfaces with high lateral resolution and wide areas. The exploitation of the vast degrees of freedom and material choice is hampered by the difficulty in the inverse design of metasurfaces. The prevalent design approach of unit-cell library creation and element juxtaposition is known to result in reduced efficiency owing to the inaccurate accounting of inter-element coupling. We report on our recent efforts in accelerated evolutionary optimization for designing metasurfaces with extended unit-cells using learned surrogate models. The difficulty in creating learned models with acceptable predictive capacity in higher dimensional parameter spaces arises from the need for extensive ground-truth generation. By a systematic study of network architectures and dataset sampling strategies, we uncover efficient ground-truth generation strategies. Specifically, we consider 2 and 3-nanoellipse titania metaatoms allowing full control over the elliptical parameters and with an optical response consisting of the spectral behavior of various transmission and reflection-mode diffracted orders for proof-of-concept demonstration. The systematic investigation reveals that densely connected neural architecture and judicious sampling strategies can allow learned model creation even with smaller ground-truth datasets.
High-transmissivity all-dielectric metasurfaces have recently attracted attention toward the realization of ultracompact optical devices and systems. Silicon-based metasurfaces, in particular, are highly promising considering the possibility of monolithic integration with complementary metal–oxide–semiconductor very large scale integration circuits. Realization of silicon-based metasurfaces operational in the visible wavelengths, however, remains a challenge. A numerical study of bilayered truncated-cone shaped nanoantenna elements is presented. Metasurfaces based on the proposed stepped conical geometry can be designed for operation in the 700- to 800-nm wavelength window and can achieve full-cycle phase response (0 to 2π) with an improved transmittance in comparison with the previously reported cylindrical geometry. A systematic parameter study of the influence of various geometrical parameters on the achievable amplitude and phase coverage is reported.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.