Exploiting Relationship for Complex-scene Image Generation
Authors: Tianyu Hua, Hongdong Zheng, Yalong Bai, Wei Zhang, Xiao-Ping Zhang, Tao Mei1584-1592
AAAI 2021 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Experimental results on Visual Genome and HICO-DET datasets show that our proposed method significantly outperforms prior arts in terms of IS and FID metrics. Based on our user study and visual inspection, our method is more effective in generating logical layout and appearance for complex-scenes. |
| Researcher Affiliation | Collaboration | 1JD AI Research 2Ryerson University |
| Pseudocode | No | The paper does not include any pseudocode or clearly labeled algorithm blocks. Procedures are described in text or via mathematical formulations. |
| Open Source Code | No | The paper does not provide an explicit statement about releasing its source code or a link to a repository for the methodology described in the paper. The only link provided (https://github.com/mseitzer/pytorch-fid) is for a third-party metric. |
| Open Datasets | Yes | The experimental results on Visual Genome (Krishna et al. 2017) and human-objects interactions dataset HICO-DET (Chao et al. 2018) demonstrate the complex-scene images generated by our proposed method follow the common sense. |
| Dataset Splits | No | The paper states: "Finally, we get 15963 train images and 4034 test images." While it provides train and test split sizes, it does not explicitly mention a separate validation dataset split or its size, which is needed to fully reproduce the data partitioning with training, validation, and testing sets. |
| Hardware Specification | No | The paper does not specify the hardware used for running the experiments. It mentions training models but provides no details on CPU, GPU, or other computing resources. |
| Software Dependencies | No | The paper mentions using "Adam" as an optimizer and models like "Inception V3" and "Res Net architecture", but it does not provide specific version numbers for any software libraries (e.g., Python, PyTorch, TensorFlow, CUDA) that would be needed for reproducibility. |
| Experiment Setup | Yes | We trained models using Adam with an initial lr=10 4 and batch size of 32 for 200 epochs. |