Deep Non-line-of-sight Imaging from Under-scanning Measurements
Authors: Yue Li, Yueyi Zhang, Juntian Ye, Feihu Xu, Zhiwei Xiong
NeurIPS 2023 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | The proposed method achieves superior performance on both synthetic data and public real-world data, as demonstrated by extensive experimental results with different under-scanning grids. |
| Researcher Affiliation | Academia | Yue Li Yueyi Zhang Juntian Ye Feihu Xu Zhiwei Xiong University of Science and Technology of China {yueli65,jt141884}@mail.ustc.edu.cn {zhyuey,feihuxu,zwxiong}@ustc.edu.cn |
| Pseudocode | No | The paper describes the architecture and components of the network in text and diagrams (Figure 2), but it does not provide any pseudocode or algorithm blocks. |
| Open Source Code | Yes | The code is available at https://github.com/Depth2World/Under-scanning_NLOS. |
| Open Datasets | Yes | For the synthetic dataset, we simulate the data using a transient renderer provided by [5], rendering 277 motorbikes from the Shape Net dataset [33]. ... For the real-world scenes, we adopt the real-world dataset provided by [14] which contains diverse objects. |
| Dataset Splits | No | The synthetic dataset is composed of 2712 training samples and 291 testing samples. The paper specifies training and testing samples but does not mention a distinct validation set or split. |
| Hardware Specification | Yes | All the experiments are conducted on a workstation equipped with 4 NVIDIA Ge Force A100 GPUs. |
| Software Dependencies | No | The proposed network is implemented in Py Torch. No specific version number for PyTorch or other software dependencies are provided. |
| Experiment Setup | Yes | The optimizer is Adam W [34] with a learning rate of 1 10 4 and a weight decay of 1 10 4. The models are trained on the synthetic dataset for 50 epochs with a batch size of 4, and directly tested on the real-world dataset, setting the high-spatial resolution to 128 128. For the loss function, the hyperparameters λ1, λ2, and λ3 is 1, 1 10 5, and 1 10 6. |