Towards Fine-Grained HBOE with Rendered Orientation Set and Laplace Smoothing
Authors: Ruisi Zhao, Mingming Li, Zheng Yang, Binbin Lin, Xiaohui Zhong, Xiaobo Ren, Deng Cai, Boxi Wu
AAAI 2024 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We validate the effectiveness of our method in the benchmarks with extensive experiments and show that our method outperforms state-of-the-art. |
| Researcher Affiliation | Collaboration | 1State Key Lab of CAD&CG, Zhejiang University 2FABU Inc 3School of Software Technology, Zhejiang University 4Ningbo Zhoushan Port Group Co.,Ltd., Ningbo, China |
| Pseudocode | No | The paper does not contain structured pseudocode or algorithm blocks. |
| Open Source Code | Yes | Project is available at: https://github.com/Whalesong-zrs/Towards Fine-grained-HBOE. |
| Open Datasets | Yes | As the largest and most valuable real-scene dataset, the MEBOW dataset contains around 130K training samples and has rich background environments. It will be used for both training and testing. Additionally, we will incorporate the RMOS dataset as supplementary training data and evaluate its value on the MEBOW test set. The data in the TUD dataset has clear and complete human body shapes and provides continuous labels. |
| Dataset Splits | No | The paper does not provide specific dataset split information (exact percentages, sample counts, citations to predefined splits, or detailed splitting methodology) for training, validation, and testing. |
| Hardware Specification | No | The paper does not provide specific hardware details (exact GPU/CPU models, processor types with speeds, memory amounts, or detailed computer specifications) used for running its experiments. |
| Software Dependencies | Yes | For the experiments in Tab. 1, we implenment these backbones based on the mmpose (Contributors 2020). |
| Experiment Setup | Yes | Input instances are cropped and resized to 256 192 while applying data augmentation techniques including flipping and scaling. For OEFormer training, we use 80 epochs with a batch size of 256 and the Adam W optimizer with initial learning rate 1 10 5. We set β to 0.2 and σ to 2.0 for the loss function. |