Deep Automatic Natural Image Matting
Authors: Jizhizi Li, Jing Zhang, Dacheng Tao
IJCAI 2021 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Results of the experiments demonstrate that our network trained on available composite matting datasets outperforms existing methods both objectively and subjectively. |
| Researcher Affiliation | Collaboration | Jizhizi Li1 , Jing Zhang1 and Dacheng Tao2 1The University of Sydney, Australia 2JD Explore Academy, JD.com, China |
| Pseudocode | No | The paper includes architectural diagrams but no structured pseudocode or algorithm blocks. |
| Open Source Code | Yes | The source code and dataset are available at https://github.com/Jizhizi Li/AIM. |
| Open Datasets | Yes | We trained our model and other representative matting models on the combination of matting datasets Comp-1k [Xu et al., 2017], HAtt [Qiao et al., 2020] and AM-2k [Li et al., 2020]. To reduce the domain gap issue, we adopted the highresolution background dataset BG-20k and the composition route RSSN proposed in [Li et al., 2020] to generate the training data... we leveraged the salient object detection dataset DUTS [Wang et al., 2017] for training since it contains more real-world images and classes. |
| Dataset Splits | No | The paper describes the datasets used for training and testing, but it does not provide specific percentages or sample counts for training, validation, and test splits needed to reproduce data partitioning. |
| Hardware Specification | Yes | We trained our model on a single NVIDIA Tesla V100 GPU with batch size as 16, Adam as optimizer. |
| Software Dependencies | No | The paper mentions several models and optimizers (e.g., Adam), but it does not specify software dependencies with version numbers (e.g., PyTorch, TensorFlow, Python versions). |
| Experiment Setup | Yes | We trained our model on a single NVIDIA Tesla V100 GPU with batch size as 16, Adam as optimizer. When pretraining on DUTS, we resized all images to 320 320, set learning rate as 1 10 4, trained for 100 epochs. During finetuning on the synthetic matting dataset, we randomly crop a patch with a size in {640 640, 960 960, 1280 1280} from each image and resized it to 320 320, set the learning rate as 1 10 6, and trained for 50 epochs. |