UAWTrack: Universal 3D Single Object Tracking in Adverse Weather
Authors: Yuxiang Yang, Hongjie Gu, Yingqi Deng, Zhekang Dong, Zhiwei He, Jing Zhang
AAAI 2025 | Venue PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Extensive experiments demonstrate that UAWTrack achieves state-of-the-art performance under all weather conditions. |
| Researcher Affiliation | Academia | 1School of Electronics and Information, Hangzhou Dianzi University, China 2School of Computer Science, Wuhan University, China EMAIL, EMAIL |
| Pseudocode | No | The paper describes the methodology using textual explanations, mathematical equations, and diagrams. There are no explicit pseudocode or algorithm blocks labeled in the document. |
| Open Source Code | Yes | Code https://github.com/HDU-VRLab/UAWTrack |
| Open Datasets | Yes | Specifically, by applying physically-based adverse weather simulation algorithms (Hahner et al. 2021, 2022; Dong et al. 2023) to two classic autonomous driving datasets, KITTI (Geiger, Lenz, and Urtasun 2012) and Nu Scenes (Caesar et al. 2020), we create two synthetic datasets: KITTI-A and Nu Scenes A. |
| Dataset Splits | Yes | Specifically, KITTI-A contains 500 sequences, which are split into training (210 sequences) and testing sets (290 sequences), following the settings in previous works (Yang et al. 2023; Xu et al. 2023b). Compared to KITTI-A, Nu Scenes-A is a more challenging dataset which contains 7,000 and 1,500 scenes for training and testing, respectively. |
| Hardware Specification | Yes | Our model is trained with a batch size of 128 and an initial learning rate of 1 10 4 using the Adam W optimizer on two NVIDIA GTX 4090 GPUs. |
| Software Dependencies | No | The network is implemented in Py Torch with MMEngine (Contributors 2022). While PyTorch and MMEngine are mentioned, specific version numbers for these software components are not provided. |
| Experiment Setup | Yes | Our model is trained with a batch size of 128 and an initial learning rate of 1 10 4 using the Adam W optimizer on two NVIDIA GTX 4090 GPUs. |