MCMinho Choi

Minho Choi

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Minho Choi has a total of 61 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    De-occluding broadband metalens

    by Seungwoo Yoon, Dohyun Kang, Eunsue Choi, Sohyun Lee, Seoyeon Kim, Minho Choi, Hyeonsu Heo, Dong-Ha Shin, Suha Kwak, Arka Majumdar, et al.

  • Meta-Optical Encoder for Image Segmentation

    by Minho Choi, Jinlin Xiang, Yubo Zhang, Zhihao Zhou, Eli Shlizerman, Arka Majumdar

    Deep neural networks achieve remarkable results in image segmentation but are often too computationally intensive for resource-limited devices. We propose a hybrid optical/digital neural network that integrates a meta-optical encoder performing convolution directly in the optical domain, drastically reducing computational complexity. Leveraging knowledge distillation, we compress a traditional U-Net into a compact parallel network and replace its initial convolutional layers with an engineered metasurface. Our approach attains an ∼12000× reduction in parameters and an ∼800× reduction in operations at the cost of an ∼16.6% accuracy loss, while maintaining an ∼81.8% segmentation accuracy. This significantly outperforms fully digital compressed models of similar complexity, which exhibit an ∼31.0% accuracy loss. This work demonstrates the first experimental optical neural network operating on incoherent light capable of complex image segmentation beyond simple classification, paving the way for practical low-power, high-speed hybrid optical/digital systems suitable for autonomous agents and edge devices.


  • Actively Tunable Metalens with Varying Fields of View

    by Hyeonsu Heo, Minho Choi, Hyunjung Kang, Kwang-Suk Oh, Seungwoo Yoon, Young-Ki Kim, Seung-Hwan Baek, Arka Majumdar, Junsuk Rho

    Metasurfaces have emerged as compelling alternatives to conventional optics, owing to their lightweight nature, compact form factor, and diverse functional capabilities. However, their lack of postfabrication tunability limits practical applications where dynamic modulation is crucial. In this work, we integrate liquid crystal and polarization-multiplexed metasurfaces to address two key functionalities: field of view and spatial resolution. We experimentally demonstrate two prototypes: (i) a two-phase switching metasurface that alternates between high-resolution narrow-view and low-resolution wide-view modes and (ii) a three-phase scanning metasurface that realizes three discrete field-of-view states separated by 2°. These results verify that reconfigurable metasurfaces can achieve multistate control of field of view and resolution without mechanical motion.


  • Top Cited Papers
    Open Access

    Transferable polychromatic optical encoder for neural networks

    by Minho Choi, Jinlin Xiang, Anna Wirth-Singh, Seung-Hwan Baek, Eli Shlizerman, Arka Majumdar

    Artificial neural networks have fundamentally transformed the field of computer vision, providing unprecedented performance. However, these neural networks for image processing demand substantial computational resources, often hindering real-time operation. In this work, we demonstrate an optical encoder that can perform convolution simultaneously in three color channels during the image capture, effectively implementing several initial convolutional layers of the network. Such an optical encoding results in ~ 24, 000 × reduction in computational operations, with a state-of-the-art classification accuracy (~73.2%) in free-space optical system. In addition, our analog optical encoder, trained for CIFAR-10 data, can be transferred to the ImageNet subset, High-10, without any modifications, and still exhibits moderate accuracy. The proposed method can decrease total system-level energy more than two orders of magnitude per a single object classification. Our results evidence the potential of hybrid optical/digital computer vision system in which the optical frontend can pre-process an ambient scene to reduce the energy and latency of the whole computer vision system.


  • Top Cited Papers

    Observation of Photonic Chiral Flatbands

    by Minho Choi, Andrea Alù, Arka Majumdar

    Distinct selectivity to the spin angular momenta of photons has garnered significant attention in recent years, for its relevance in basic science and for imaging and sensing applications. While nonlocal metasurfaces with strong chiral responses to the incident light have been reported, these responses are typically limited to a narrow range of incident angles. In this study, we demonstrate a nonlocal metasurface that showcases strong chirality, characterized by circular dichroism (∼0.6), over a wide range of incident angles ±5°. Its quality factor, circular dichroism and resonant frequency can be optimized by design. These findings pave the way to further advance the development of valley-selective optical cavities and augmented reality applications.


  • Open Access

    Compressed meta-optical encoder for image classification

    by Anna Wirth-Singh, Jinlin Xiang, Minho Choi, Johannes E. Fröch, Luocheng Huang, Shane Colburn, Eli Shlizerman, Arka Majumdar

    Convolutional neural networks (CNNs) represent a significant milestone in image classification, recognition, and tracking.1 CNNs, for example, AlexNet, are composed of several convolutional layers that adaptively learn spatial representations from input images. Although powerful, the convolution operation is computationally expensive, leading to high latency and power consumption. In fact, it has been estimated that ∼80% of the total runtime of CNNs is used in performing convolution operations.2 Reducing this latency and, as a result, power consumption has become an active area of research, with multiple works proposing free-space optical systems as a solution.3 – 8 Beyond reducing latency and power consumption, optical information processing offers advantages including high bandwidth, spatial parallelism, and low-loss transmission, which have led to a surge of interest in the field.3


  • Open Access

    Transferable polychromatic optical encoder for neural networks

    by Arka Majumdar, Minho Choi, Jinlin Xiang, Anna Wirth-Singh, Seung-Hwan Baek, Eli Shlizerman

    Artificial neural networks (ANNs) have fundamentally transformed the field of computer vision, providing unprecedented performance. However, these ANNs for image processing demand substantial computational resources, often hindering real-time operation. In this paper, we demonstrate an optical encoder that can perform convolution simultaneously in three color channels during the image capture, effectively implementing several initial convolutional layers of a ANN. Such an optical encoding results in ~24,000 times reduction in computational operations, with a state-of-the art classification accuracy (~ 73.2%) in free-space optical system. In addition, our analog optical encoder, trained for CIFAR-10 data, can be transferred to the ImageNet subset, High-10, without any modifications, and still exhibits moderate accuracy. Our results evidence the potential of hybrid optical/digital computer vision system in which the optical frontend can pre-process an ambient scene to reduce the energy and latency of the whole computer vision system.

  • Meta-optics triplet for zoom imaging at mid-wave infrared

    by Anna Wirth-Singh, Arturo Martin Jimenez, Minho Choi, Johannes E. Fröch, Rose Johnson, Tina Le Teichmann, Zachary Coppens, Arka Majumdar

    Lenses with dynamic focal length, also called zoom functionality, enable a variety of applications related to imaging and sensing. The traditional approach of stacking refractive lenses to achieve this functionality results in an expensive, heavy optical system. Especially for applications in the mid-infrared, light weight and compact form factor are required. In this work, we use a meta-optic triplet to demonstrate zoom imaging at mid-wave infrared wavelengths. By varying the axial distances between the optics, the meta-optic triplet achieves high-quality imaging over a zoom range of 5×, with a 50° full field of view in the widest configuration and an aperture of 8 mm. This triplet system demonstrates the potential for meta-optics to reduce conventional components in complex and multi-functional imaging systems to dramatically thinner and lighter components.


  • Open Access

    Transferable polychromatic optical encoder for neural networks

    by Minho Choi, Jinlin Xiang, Anna Wirth-Singh, Seung-Hwan Baek, Eli Shlizerman, Arka Majumdar

    Artificial neural networks (ANNs) have fundamentally transformed the field of computer vision, providing unprecedented performance. However, these ANNs for image processing demand substantial computational resources, often hindering real-time operation. In this paper, we demonstrate an optical encoder that can perform convolution simultaneously in three color channels during the image capture, effectively implementing several initial convolutional layers of a ANN. Such an optical encoding results in ~24,000 times reduction in computational operations, with a state-of-the art classification accuracy (~73.2%) in free-space optical system. In addition, our analog optical encoder, trained for CIFAR-10 data, can be transferred to the ImageNet subset, High-10, without any modifications, and still exhibits moderate accuracy. Our results evidence the potential of hybrid optical/digital computer vision system in which the optical frontend can pre-process an ambient scene to reduce the energy and latency of the whole computer vision system.


  • Open Access

    Observation of Photonic Chiral Flatbands

    by Minho Choi, Andrea Alù, Arka Majumdar

    Distinct selectivity to the spin angular momenta of photons have garnered significant attention in recent years, for their relevance in basic science and for imaging and sensing applications. While nonlocal metasurfaces with strong chiral responses to the incident light have been reported, these responses are typically limited to a narrow range of incident angles. In this study, we demonstrate a nonlocal metasurface that showcases strong chirality, characterized by circular dichroism larger than 0.6, over a wide range of incident angles $\pm5^o$. Its quality factor, circular dichroism and resonant frequency can be optimized by design. These findings pave the way to further advance the development of valley-selective optical cavities and augmented reality applications.