Saved in:
Bibliographic Details
Main Authors: Nguyen, Pham Phuong Nam, Le, Nam Tien, Vo, Thi Kim Trang, Nguyen, Nhu Tinh Anh
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2604.20291
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866914498333376512
author Nguyen, Pham Phuong Nam
Le, Nam Tien
Vo, Thi Kim Trang
Nguyen, Nhu Tinh Anh
author_facet Nguyen, Pham Phuong Nam
Le, Nam Tien
Vo, Thi Kim Trang
Nguyen, Nhu Tinh Anh
contents Efficient single-image super-resolution (SISR) requires balancing reconstruction fidelity, model compactness, and robustness under low-bit deployment, which is especially challenging for x3 SR. We present a deployment-oriented quantized SISR framework based on an extract-refine-upsample design. The student performs most computation in the low-resolution space and uses a lightweight re-parameterizable backbone with PixelShuffle reconstruction, yielding a compact inference graph. To improve quality without significantly increasing complexity, we adopt a three-stage training pipeline: Stage 1 learns a basic reconstruction mapping with spatial supervision; Stage 2 refines fidelity using Charbonnier loss, DCT-domain supervision, and confidence-weighted output-level distillation from a Mamba-based teacher; and Stage 3 applies quantization-aware training directly on the fused deploy graph. We further use weight clipping and BatchNorm recalibration to improve quantization stability. On the MAI 2026 Quantized 4K Image Super-Resolution Challenge test set, our final AIO MAI submission achieves 29.79 dB PSNR and 0.8634 SSIM, obtaining a final score of 1.8 under the target mobile INT8 deployment setting. Ablation on Stage 3 optimization shows that teacher-guided supervision improves the dynamic INT8 TFLite reconstruction from 29.91 dB/0.853 to 30.0003 dB/0.856, while the fixed-shape deployable INT8 TFLite artifact attains 30.006 dB/0.857.
format Preprint
id arxiv_https___arxiv_org_abs_2604_20291
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Efficient INT8 Single-Image Super-Resolution via Deployment-Aware Quantization and Teacher-Guided Training
Nguyen, Pham Phuong Nam
Le, Nam Tien
Vo, Thi Kim Trang
Nguyen, Nhu Tinh Anh
Computer Vision and Pattern Recognition
Efficient single-image super-resolution (SISR) requires balancing reconstruction fidelity, model compactness, and robustness under low-bit deployment, which is especially challenging for x3 SR. We present a deployment-oriented quantized SISR framework based on an extract-refine-upsample design. The student performs most computation in the low-resolution space and uses a lightweight re-parameterizable backbone with PixelShuffle reconstruction, yielding a compact inference graph. To improve quality without significantly increasing complexity, we adopt a three-stage training pipeline: Stage 1 learns a basic reconstruction mapping with spatial supervision; Stage 2 refines fidelity using Charbonnier loss, DCT-domain supervision, and confidence-weighted output-level distillation from a Mamba-based teacher; and Stage 3 applies quantization-aware training directly on the fused deploy graph. We further use weight clipping and BatchNorm recalibration to improve quantization stability. On the MAI 2026 Quantized 4K Image Super-Resolution Challenge test set, our final AIO MAI submission achieves 29.79 dB PSNR and 0.8634 SSIM, obtaining a final score of 1.8 under the target mobile INT8 deployment setting. Ablation on Stage 3 optimization shows that teacher-guided supervision improves the dynamic INT8 TFLite reconstruction from 29.91 dB/0.853 to 30.0003 dB/0.856, while the fixed-shape deployable INT8 TFLite artifact attains 30.006 dB/0.857.
title Efficient INT8 Single-Image Super-Resolution via Deployment-Aware Quantization and Teacher-Guided Training
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2604.20291