license: apache-2.0
Palmpad Dataset
The Palmpad Dataset is a publicly available multimodal dataset designed for research in finger-to-palm recognition. This dataset contains data from users, divided into two groups (Group 1 and Group 2). Each user's data consists of multiple video files in .avi format, along with corresponding frame-level annotation files in .txt format.
Overview
Paper Title: Palmpad: Enabling Real-Time Index-to-Palm Touch Interaction with a Single RGB Camera
Dataset Content
Each data sample includes a .avi video file and a corresponding .txt annotation file with frame-level labels. The dataset is divided into two groups. Applications: Finger-to-Palm Recognition
Dataset Structure
The dataset is organized as follows:
Palmpad_Dataset/ │ ├── group_1/ │ ├── user-0/ │ │ ├── user-0_0.avi │ │ ├── user-0_0.txt │ │ ├── user-0_1.avi │ │ ├── user-0_1.txt │ │ └── ... │ ├── user-1/ │ │ └── ... │ └── ... │ ├── group_2/ │ ├── user-0/ │ │ ├── user-0_0.avi │ │ ├── user-0_0.txt │ │ └── ... │ └── ... └── ...
File Descriptions group_1/ and group_2/: The dataset is split into two groups, each containing data from 16 users. user-{id}/: Data for each user is stored in a separate folder. user-{id}{index}.avi: Video file capturing the user's action. user-{id}{index}.txt: Frame-level annotation file corresponding to the video.
Annotation Format The annotation files are plain text, where each line corresponds to a frame in the video. The format is as follows: : The frame timestamp for each frame. : The label of touch(1) or not(0) for the corresponding frame. : Middle value for calculate the label.
Example: 1691995008118219 0 0 1691995008126552 0 0 1691995008134885 0 0
Dataset Statistics
User number: 16 users in Group 1, 16 users in Group 2. Average Data per User: Each user has multiple video files and corresponding annotation files. Video Resolution: 1280x720 Video Frame Rate: 120 FPS
License
This dataset is released under the following license: Apache 2.0 For commercial or other uses, please contact the authors for permission.
Citation
If you use the Palmpad Dataset in your research, please cite the following paper: Zhe He, Xiangyang Wang, Yuanchun Shi, Chi Hsia, Chen Liang, and Chun Yu. 2025. Palmpad: Enabling Real-Time Index-to-Palm Touch Interaction with a Single RGB Camera. In Proceedings of the 2025 CHI Conference on Human Factors in Computing Systems (CHI '25). Association for Computing Machinery, New York, NY, USA, Article 551, 1–16. https://doi.org/10.1145/3706598.3714130
Contact
If you have any questions or suggestions, please contact us via: Email: [email protected] Issues: Submit an issue on the Issues page of this repository.