{"ID":2844558,"CreatedAt":"2026-06-01T04:54:23.091178241Z","UpdatedAt":"2026-06-01T04:54:23.091178241Z","DeletedAt":null,"paper_url":"https://arxiv.org/abs/2511.05868","arxiv_id":"2511.05868","title":"HarmoQ: Harmonized Post-Training Quantization for High-Fidelity Image","abstract":"Post-training quantization offers an efficient pathway to deploy super-resolution models, yet existing methods treat weight and activation quantization independently, missing their critical interplay. Through controlled experiments on SwinIR, we uncover a striking asymmetry: weight quantization primarily degrades structural similarity, while activation quantization disproportionately affects pixel-level accuracy. This stems from their distinct roles--weights encode learned restoration priors for textures and edges, whereas activations carry input-specific intensity information. Building on this insight, we propose HarmoQ, a unified framework that harmonizes quantization across components through three synergistic steps: structural residual calibration proactively adjusts weights to compensate for activation-induced detail loss, harmonized scale optimization analytically balances quantization difficulty via closed-form solutions, and adaptive boundary refinement iteratively maintains this balance during optimization. Experiments show HarmoQ achieves substantial gains under aggressive compression, outperforming prior art by 0.46 dB on Set5 at 2-bit while delivering 3.2x speedup and 4x memory reduction on A100 GPUs. This work provides the first systematic analysis of weight-activation coupling in super-resolution quantization and establishes a principled solution for efficient high-quality image restoration.","short_abstract":"Post-training quantization offers an efficient pathway to deploy super-resolution models, yet existing methods treat weight and activation quantization independently, missing their critical interplay. Through controlled experiments on SwinIR, we uncover a striking asymmetry: weight quantization primarily degrades struc...","url_abs":"https://arxiv.org/abs/2511.05868","url_pdf":"https://arxiv.org/pdf/2511.05868v2","authors":"[\"Hongjun Wang\",\"Jiyuan Chen\",\"Xuan Song\",\"Yinqiang Zheng\"]","published":"2025-11-08T05:53:47Z","proceeding":"eess.IV","tasks":"[\"eess.IV\",\"cs.CV\"]","methods":"[]","has_code":false}