HMHyeon-Cheol Moon

Hyeon-Cheol Moon

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

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Hyeon-Cheol Moon has a total of 15 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • GQKD: Greedy Query Knowledge Distillation for Efficient 3D Instance Segmentation

    by Taehyun Jeong, Hyeon-Cheol Moon, Suhyeon Lee, Jinwoo Jeong

    3D instance segmentation is a fundamental component of various 3D scene understanding applications. Recent transformer-based approaches have achieved strong performance through query-driven decoding. however, their repeated multi-layer decoder architectures impose substantial computational overhead, motivating the need for model compression. In this work, we identify two key phenomena across decoder layers: object disappearance, where valid object queries vanish as decoding proceeds, and object degradation, where intermediate-layer predictions can be more reliable than final-layer outputs. Based on these observations, we propose Greedy Query Knowledge Distillation (GQKD), an effective distillation framework for lightweight 3D instance segmentation. GQKD constructs a pseudo-GT query set to supervise the student model by selecting the best true-positive query for each instance across multiple decoder layers, along with low-confidence negative teacher queries. This Pseudo-GT recovers high-quality object hypotheses lost during decoding and provides balanced supervision by supplementing recovered positives with low-confidence negatives. Experimental results on ScanNetV2 demonstrate that GQKD enables a student model with approximately 50% reduced decoder depth to achieve performance close to the teacher when used alone, and further improves results when combined with existing distillation strategies.

  • Open Access

    Feature-Domain Adaptive Contrastive Distillation for Efficient Single Image Super-Resolution

    by Hyeon-Cheol Moon, Jae-Gon Kim, Jinwoo Jeong, Sungjei Kim

    Convolutional neural network-based single image super-resolution (SISR) involves numerous parameters and high computational expenses to ensure improved performance, limiting its applicability in resource-constrained devices such as mobile phones. Knowledge distillation (KD), which transfers useful knowledge from a teacher network to a student network, has been investigated as a method to make networks more efficient in terms of performance. To this end, feature distillation (FD) has been utilized in KD to minimize the Euclidean distance-based loss of feature maps between teacher and student networks. However, this technique does not adequately consider the effective and meaningful delivery of knowledge from the teacher to the student network to improve the latter’s performance under given network capacity constraints. In this study, we propose a feature-domain adaptive contrastive distillation (FACD) method to train lightweight student SISR networks efficiently. We highlight the limitations of existing FD methods in terms of Euclidean distance-based loss, and propose a feature-domain contrastive loss, which causes student networks to learn richer information from the teacher’s representation in the feature domain. We also implement adaptive distillation that performs distillation selectively depending on the conditions of the training patches. Experimental results showed that the proposed FACD scheme improves student enhanced deep residual networks and residual channel attention networks not only in terms of the peak signal-to-noise ratio on all benchmark datasets and scales but also in terms of subjective image quality, compared to the conventional FD approaches. Code will be release at https://github.com/hcmoon0613/FACD.


  • Open Access

    Skip-Concatenated Image Super-Resolution Network for Mobile Devices

    by Ganzorig Gankhuyag, Jingang Huh, Myeongkyun Kim, Kihwan Yoon, Hyeoncheol Moon, Seungho Lee, Jinwoo Jeong, Sungjei Kim, Yoonsik Choe

    Single-image super-resolution technology has been widely studied in various applications to improve the quality and resolution of degraded images acquired from noise-sensitive low-resolution sensors. As most studies on single-image super-resolution focused on the development of deep learning networks operating on high-performance GPUs, this study proposed an efficient and lightweight super-resolution network that enables real-time performance on mobile devices. To replace the relatively slow element-wise addition layer on mobile devices, we introduced a skip connection layer by directly concatenating a low-resolution input image with an intermediate feature map. In addition, we introduced weighted clipping to reduce the quantization errors commonly encountered during float-to-int8 model conversion. Moreover, a reparameterization method was selectively applied without increasing the cost in terms of inference time and number of parameters. Based on the contributions, the proposed network has been recognized as the best solution in Mobile AI & AIM 2022 Real-Time Single-Image Super-Resolution Challenge with PSNR of 30.03 dB and NPU runtime of 19.20 ms.