Semantic-Guided Diffusion Model for Single-Step Image Super-Resolution

Authors: Zihang Liu, Zhenyu Zhang, Hao Tang

IJCAI 2025 | Venue PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Extensive experiments on both real-world and synthetic datasets demonstrate that SAMSR significantly improves perceptual quality and detail recovery, particularly in semantically complex images.
Researcher Affiliation Academia 1Beijing Institute of Technology 2Nanjing University 3School of Computer Science, Peking University EMAIL, EMAIL, EMAIL
Pseudocode Yes Algorithm 1 Training the Pixel-wise Sampling Framework
Open Source Code Yes Our code is released at https://github.com/Liu-Zihang/SAMSR.
Open Datasets Yes For real-world evaluation, we utilize Real SR [Cai et al., 2019b] and Real Set65 [Yue et al., 2024]. For synthetic datasets, we follow the standard pipeline to create LR inputs from 3000 HR images randomly selected from Image Net [Wang et al., 2024].
Dataset Splits No For synthetic datasets, we follow the standard pipeline to create LR inputs from 3000 HR images randomly selected from Image Net [Wang et al., 2024]. The paper does not explicitly provide details on how these images were split into training, validation, or test sets for their experiments.
Hardware Specification Yes Table 5: A comparison of the training time cost and results on NVIDIA RTX4090.
Software Dependencies No The paper does not explicitly state specific software dependencies with version numbers.
Experiment Setup Yes Specifically, our model achieves convergence in only 10,000-15,000 iterations. The hyper-parameter m controls the noise addition speed and intensity for pixels with different levels of semantic richness during the forward diffusion process. ... Therefore, in this paper, we set m to 1/5. ... where ΜΈ is a hyper-parameter that controls the contribution of the semantic consistency loss.