Accelerating Diffusion-based Super-Resolution with Dynamic Time-Spatial Sampling
Authors: Rui Qin, Qijie Wang, Ming Sun, Haowei Zhu, Chao Zhou, Bin Wang
IJCAI 2025 | Venue PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Extensive evaluations across multiple benchmarks demonstrate that TSS achieves state-of-the-art (SOTA) performance with significantly fewer iterations, improving MUSIQ scores by 0.2 3.0 and outperforming the current acceleration methods with only half the number of steps. |
| Researcher Affiliation | Collaboration | 1School of Software, Tsinghua University, Beijing, China 2Kuaishou Technology, Beijing, China |
| Pseudocode | No | The paper describes the proposed Time Dynamic Sampling (TDS) and Spatial Dynamic Sampling (SDS) strategies using mathematical equations and descriptive text (e.g., equations 1, 2, 3, 4, 5, 6, 7) and illustrative figures. However, it does not contain a dedicated section or block explicitly labeled as 'Pseudocode' or 'Algorithm' with structured steps. |
| Open Source Code | No | The paper states: '1The full version is available at https://arxiv.org/abs/2505.12048.' This link is for the paper itself and not for the source code. There is no explicit statement about releasing code for the methodology described in the paper. |
| Open Datasets | Yes | Testing Datasets Performance was evaluated on both synthetic and real-world data. Synthetic data used DIV2K [2017] with LR images degraded by BSRGAN [2021]. Real-world evaluations used DReal SR [2020], Real Photo60 [2024], and Real SR [2020], along with face SR benchmarks Web Photo Test [2021a] and LFW-test [2019]. |
| Dataset Splits | No | The paper mentions several datasets (DIV2K, Real SR, DReal SR, Real Photo60, Web Photo Test, LFW-test) for evaluation and states the scale factors used (x2 or x4). It also mentions that 'Real Photo60 images were divided into non-overlapping 128 × 128 patches'. However, it does not provide specific training/test/validation split percentages, sample counts, or explicit references to predefined splits for general experimental reproduction. |
| Hardware Specification | Yes | Experiments were conducted on an NVIDIA A800 GPU (80GB) using the official codebases and pre-trained weights. |
| Software Dependencies | No | The PyTorch framework was used for implementation. However, no specific version number for PyTorch or any other software dependency is provided. |
| Experiment Setup | No | The paper states: 'The detailed hyperparameters are provided in the full version.' This indicates that the specific experimental setup details, including hyperparameters, are not present in the main body of the paper. |