Dr. Daniel ROGGEN led this dataset project, bringing in his expertise in creating large scale activity recognition datasets for the scientific community. He worked on multiple parts of this project, including obtaining ethical approvals, hiring participants, defining the data collection and annotation protocols and eventually defining the dissemination strategy. Technically, he set-up the infrastructure to backup and share the dataset among team members. He developed the quality control scripts and the dataset curation pipeline, bringing 1 terabyte of multimodal data into a manageable format. Prior to this, he produced other recognised activity recognition datasets, such as the Skoda and OPPORTUNITY datasets of human activities.
Dr. Stefan VALENTIN is a principal researcher and team leader at Huawei’s Research Center in Paris, France. He is excited to understand how accurately we can estimate and predict user mobility based on Smartphone sensors. By classifying how users move, this project provides an important building block to answer this question. Stefan was responsible for initiating the project and is supervising it now as technical project manager.
Dr. Lin WANG is responsible for data analysis and activity recognition. He applies signal processing approaches to analyze the temporal and frequency characteristics of the dataset. He develops reference machine leaning algorithms, including novel deep learning algorithms, to recognize the transportation modes of the users from the multimodal sensor data of the mobile phones. He places a specific emphasis on multimodal adaptive sensor fusion.
Dr. Hristijan GJORESKI set up the data collection protocol and supervised the participants performing data collection. He planned the activity scenarios on a weekly basis, and performed a qualitative data quality check-up continuously during the collection of the data. He was also included in the initial data analysis and setting up the activity recognition pipeline.
Mathias CILIBERTO developed the toolchain to manage the raw data and to set-up the post-hoc video-based re-annotation. He developed scripts to automate most of the process; from downloading the data from the phones to organising the data hierachically, and checking the data quality. Moreover, he synchronized the data with the body worn camera and set-up the system to allow participants to verify and correct their annotation. Finally, he provided support for the Android data logging application, adjusting it according with the requirements during the data collection.
Dr. Sami MEKKI is senior researcher at HUAWEI research center in Paris, France, within Mathematical and Algorithmic Sciences Lab. Sami is interested in sensor data fusion for user tracking and localization. This will help to get a better estimation of the propagation channel and radio resource optimization.
Francisco Javier Ordoñez Morales
Dr. Javier ORDOÑEZ MORALES was involved in the first steps of the project, shaping its scope and defining the work packages. He helped designing the data collection and annotation protocols, and the overall experimental setup. He developed and tested the Android data logging application used to collect the data.