finemotion
FineMotion: A Dataset and Benchmark with both Spatial and Temporal Annotation for Fine-grained Motion Generation and Editing
Science Score: 41.0%
This score indicates how likely this project is to be science-related based on various indicators:
-
✓CITATION.cff file
Found CITATION.cff file -
✓codemeta.json file
Found codemeta.json file -
○.zenodo.json file
-
○DOI references
-
✓Academic publication links
Links to: arxiv.org -
○Academic email domains
-
○Institutional organization owner
-
○JOSS paper metadata
-
○Scientific vocabulary similarity
Low similarity (10.6%) to scientific vocabulary
Repository
FineMotion: A Dataset and Benchmark with both Spatial and Temporal Annotation for Fine-grained Motion Generation and Editing
Basic Info
- Host: GitHub
- Owner: BizhuWu
- Default Branch: main
- Size: 16.8 MB
Statistics
- Stars: 4
- Watchers: 1
- Forks: 0
- Open Issues: 0
- Releases: 0
Metadata Files
README.md
FineMotion: A Dataset and Benchmark with both Spatial and Temporal Annotation for Fine-grained Motion Generation and Editing
FineMotion provides strictly aligned, fine-grained descriptions of human body part movements (BPM) for both motion snippets (short segments of motion sequences) and entire motion sequences, with motions sourced from HumanML3D).
Statistics of FineMotion
** ### :bar_chart: Statistics In summary, our FineMotion dataset contains **21,346** human-annotated BPM descriptions and **420,968** automatically generated ones for diverse motion snippets, i.e., BPMSD. Notably, the temporal information within the textual annotations facilitates easy augmentation, such as performing random cropping along the temporal dimension. This approach can generate numerous pairs of motion clips—each consisting of several adjacent snippets—along with their corresponding BPM descriptions. Additionally, the dataset includes **4,492** BPM paragraphs (BPMP) organized from human-annotated BPM snippet descriptions, and **89,940** paragraphs organized from automatically generated BPM snippet descriptions, covering a total of 29,232 motion sequences. For BPMSD, the average and median lengths are **18** and **19** words, respectively. As for BPMP, the average and median lengths are **244** and **220** words, respectively. ### :woman_judge: Data Quality TODOIf this dataset is useful in your projects, we will apprecite your star on this codebase. 😆😆
Checkout Works on FineMotion
📌 (CVPR 2025) MG-MotionLLM: A Unified Framework for Motion Comprehension and Generation across Multiple Granularities [paper][code]
How to Obtain the Data
For motion sequences, you could find the details as well as download link here).
For our fine-grained textual descriptions, you could directly download them from this repository under the folder "./FineMotion".
Data Structure
sh
<FineMotion>
./BPMP_auto.zip // Automatically-Generated Body Part Movement Paragraph.
./BPMSD_auto.zip // Automatically-Generated Body Part Snippet Descriptions.
./BPMP_human.zip // Human-Annotated Body Part Movement Paragraph. Coming Soon!
./BPMSD_human.zip // Human-Annotated Body Part Snippet Description. Coming Soon!
Unzip all files under the folder "./FineMotion" to obtain four JSON files: 'BPMPauto.json', 'BPMSDauto.json', 'BPMPhuman.json', and 'BPMSDhuman.json'. All JSON files use a dictionary format. The keys follow the same naming protocols as HumanML3D for consistency, while the values contain fine-grained textual descriptions of the corresponding motion sequences in HumanML3D. Specifically, the description in the value of the key "XXXXXX" (e.g., '000000') precisely describes the human motion stored in "./newjoints (or newjoint_vecs)/XXXXXX.npy" (e.g., '000000.npy') in HumanML3D.
An example of all the detailed human body part snippet descriptions (BPMSDs) in a whole human motion sequence is shown below:
sh
{
"000314": # name of motion sequence
[
"", # 0.0s-0.5s’ BPMSD
"Raise your hands up to your head. Turn your head to the left.”,
"", # 1.0s-1.5s’ BPMSD
"", # 1.5s-2.0s’ BPMSD
"Lean to the right. Move your hands closer together. Turn your head to the left.",
"", # 2.5s-3.0s’ BPMSD
"Lower your arms to your sides. Turn your head to the left.",
"Lean your upper body to the right. Move your right hand to your right thigh. Move your left hand to your left thigh."
],
}
Each item in the list (i.e. the value of the key "000314") represents a fine-grained human body movement snippet description for a 0.5s interval.
An example for three different detailed human body part paragraphs (BPMPs) for the same human motion sequence is shown below:
sh
{
"000314": # name of motion sequence
[
"Beginning with his hands raised overhead and his head turned left, the individual leans to the right while bringing his hands closer and maintaining his head position. Continuing the movement, he lowers his arms to his sides while keeping his head turned left. Subsequently, he leans his upper body to the right and simultaneously moves his right hand to his right thigh and his left hand to his left thigh.",
"Initially, the individual raises both hands above his head and turns his head to the left. He then leans to the right while bringing his hands closer together. Simultaneously, he maintains his head turned to the left. Subsequently, he lowers his arms to his sides and continues turning his head to the left. Finally, he leans his upper body to the right, placing his right hand on his right thigh and his left hand on his left thigh.",
”Initially, the individual raises his hands to his head and turns his head to the left. Subsequently, he leans to the right while continuing to move his hands closer together and turning his head to the left. Next, he lowers his arms to his sides and keeps turning his head to the left. Finally, he leans his upper body to the right, placing his right hand on his right thigh and his left hand on his left thigh."
],
}
Each item in the list (i.e. the value of key "000314") represents a fine-grained human body movement paragraph for the entire motion sequence.
Citation
If you are using FineMotion dataset, please consider citing this repository:
@misc{Wu_FineMotion_Dataset,
author = {Wu, Bizhu and Xie, Jinheng and Ding, Meidan and Kong, Zhe and Ren, Jianfeng and Bai, Ruibin and Qu, Rong and Shen, Linlin},
title = {{FineMotion Dataset}},
url = {https://github.com/BizhuWu/FineMotion}
}
Acknowledgments
Thanks to HumanML3D, our repository is partially borrowing from them.
Owner
- Name: BizhuWu
- Login: BizhuWu
- Kind: user
- Repositories: 2
- Profile: https://github.com/BizhuWu
Citation (CITATION.cff)
type: dataset cff-version: 1.2.0 message: "If you use this dataset, please cite it as below." authors: - family-names: "Wu" given-names: "Bizhu" - family-names: "Xie" given-names: "Jinheng" - family-names: "Ding" given-names: "Meidan" - family-names: "Kong" given-names: "Zhe" - family-names: "Ren" given-names: "Jianfeng" - family-names: "Bai" given-names: "Ruibin" - family-names: "Qu" given-names: "Rong" - family-names: "Shen" given-names: "Linlin" title: "FineMotion Dataset" version: 0.0.1 doi: date-released: url: "https://github.com/BizhuWu/FineMotion"
GitHub Events
Total
- Watch event: 21
- Push event: 1
- Public event: 1
Last Year
- Watch event: 21
- Push event: 1
- Public event: 1
Issues and Pull Requests
Last synced: over 1 year ago
All Time
- Total issues: 0
- Total pull requests: 0
- Average time to close issues: N/A
- Average time to close pull requests: N/A
- Total issue authors: 0
- Total pull request authors: 0
- Average comments per issue: 0
- Average comments per pull request: 0
- Merged pull requests: 0
- Bot issues: 0
- Bot pull requests: 0
Past Year
- Issues: 0
- Pull requests: 0
- Average time to close issues: N/A
- Average time to close pull requests: N/A
- Issue authors: 0
- Pull request authors: 0
- Average comments per issue: 0
- Average comments per pull request: 0
- Merged pull requests: 0
- Bot issues: 0
- Bot pull requests: 0