Text me the data: Generating Ground Pressure Sequence from Textual Descriptions for HAR

被引:2
|
作者
Ray, Lala Shakti Swarup [1 ]
Zhou, Bo [1 ,2 ]
Suh, Sungho [1 ,2 ]
Krupp, Lars [1 ,2 ]
Rey, Vitor Fortes [1 ,2 ]
Lukowicz, Paul [1 ,2 ]
机构
[1] DFKI, Kaiserslautern, Germany
[2] RPTU Kaiserslautern Landau, Kaiserslautern, Germany
关键词
HAR; Generative model; GPT; Pressure Sensor; Synthetic dataset;
D O I
10.1109/PerComWorkshops59983.2024.10503379
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
In human activity recognition (HAR), the availability of substantial ground truth is necessary for training efficient models. However, acquiring ground pressure data through physical sensors itself can be cost-prohibitive, time-consuming. To address this critical need, we introduce Text-to-Pressure (T2P), a framework designed to generate extensive ground pressure sequences from textual descriptions of human activities using deep learning techniques. We show that the combination of vector quantization of sensor data along with simple text conditioned auto regressive strategy allows us to obtain high-quality generated pressure sequences from textual descriptions with the help of discrete latent correlation between text and pressure maps. We achieved comparable performance on the consistency between text and generated motion with an R squared value of 0.722, Masked R squared value of 0.892, and FID score of 1.83. Additionally, we trained a HAR model with the the synthesized data and evaluated it on pressure dynamics collected by a real pressure sensor which is on par with a model trained on only real data. Combining both real and synthesized training data increases the overall macro F1 score by 5.9 percent.
引用
收藏
页码:461 / 464
页数:4
相关论文
共 50 条
  • [31] Improving knowledge of cyclist crashes based on hospital data including crash descriptions from open text fields
    Moller, Mette
    Janstrup, Kira Hyldekaer
    Pilegaard, Ninette
    JOURNAL OF SAFETY RESEARCH, 2021, 76 : 36 - 43
  • [32] Automated extraction of data from text using an XML parser: An earth science example using fossil descriptions
    Curry, Gordon B.
    Connor, Richard C. H.
    GEOSPHERE, 2008, 4 (01): : 159 - 169
  • [33] Extracting Process Graphs from Medical Text Data An Approach Towards a Systematic Framework to Extract and Mine Medical Sequential Processes Descriptions from Large Text Sources
    Niekler, Andreas
    Kahmann, Christian
    KNOWLEDGE GRAPHS AND LANGUAGE TECHNOLOGY, 2017, 10579 : 76 - 88
  • [35] Variability of ground motions in southern California - Data from the 1995 to 1996 Ridgecrest sequence
    Scrivner, CW
    Helmberger, DV
    BULLETIN OF THE SEISMOLOGICAL SOCIETY OF AMERICA, 1999, 89 (03) : 626 - 639
  • [36] Generating DNA sequence data with limited resources formolecular biology: Lessons from a barcoding project in Indonesia
    Dean, Gillian H.
    Asmarayani, Rani
    Ardiyani, Marlina
    Santika, Yessi
    Triono, Teguh
    Mathews, Sarah
    Webb, Campbell O.
    APPLICATIONS IN PLANT SCIENCES, 2018, 6 (07):
  • [37] Exploratory analysis of textual data from the Mother and Child Handbook using a text mining method (II): Monthly changes in the words recorded by mothers
    Tagawa, Miki
    Matsuda, Yoshio
    Manaka, Tomoko
    Kobayashi, Makiko
    Ohwada, Michitaka
    Matsubara, Shigeki
    JOURNAL OF OBSTETRICS AND GYNAECOLOGY RESEARCH, 2017, 43 (01) : 100 - 105
  • [38] A history of Middle East (ME) populations from a viewpoint of the integrated whole genome sequence data and its application to identification of ME - specific variant alleles for diabetes
    Mineta, Katsuhiko
    Goto, Kosuke
    Gojobori, Takashi
    HUMAN GENOMICS, 2018, 12
  • [39] Efficient text mining method and simple tweaks for discovering and updating unknown foreign words and improving association rules extraction from textual data
    Khan, Irfan Ajmal
    Seo, Ji-Hoon
    Choi, Jin-Tak
    ASIA LIFE SCIENCES, 2015, : 663 - 680
  • [40] Efficient text mining method and simple tweaks for discovering and updating unknown foreign words and improving association rules extraction from textual data
    Khan, Irfan Ajmal
    Choi, Jin-Tak
    ASIA LIFE SCIENCES, 2015, : 71 - 88