SyFormer: Structure-Guided Synergism Transformer for Large-Portion Image Inpainting

Authors: Jie Wu, Yuchao Feng, Honghui Xu, Chuanmeng Zhu, Jianwei Zheng

AAAI 2024 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Extensive experiments are conducted on two publicly available datasets, i.e., Celeb A-HQ and Places2, to qualitatively and quantitatively demonstrate the superiority of our model over state-of-the-arts.
Researcher Affiliation Academia 1Zhejiang University of Technology 2Zhejiang University
Pseudocode No The paper does not contain any pseudocode or clearly labeled algorithm blocks.
Open Source Code No The paper does not provide any concrete access to source code for the methodology described.
Open Datasets Yes Two well-known datasets, i.e., Celeb A-HQ (Karras et al. 2018) and Place2 (Zhou et al. 2017), are used for the performance investigation.
Dataset Splits Yes The Celeb A-HQ data is split into training, validation, and test sets in a ratio of 24:1:5. We keep 220,000 and 5000 images from the original places2 sets for training and testing, respectively.
Hardware Specification Yes All experiments are conducted on two GPUs of RTX 3090 with a single 12G of video memory.
Software Dependencies No The paper mentions "Py Torch" but does not specify a version number or other software dependencies with version numbers.
Experiment Setup Yes All experiments are conducted using Py Torch with a batch size of 8. Our model is optimized by Adam with a learning rate of 2 10 4. The hyper-parameters in Eq. (15) are set as λadv = 0.1, λrec = 40, λsty = 120, λper = 0.05 to generate the sensuously optimal results.