Test dataset release:
15 Dec, 2018 03 Jan, 2019
15 Dec, 2018 03 Jan, 2019
Prediction results and technical reports submission deadline:
1 Feb, 2019 5 Feb, 2019
2 Feb, 2019 6 Feb, 2019
Dr. Trung Thanh Ngo, Osaka University
Prof. Md Atiqur Rahman Ahad, Osaka University
Assoc. Prof. Daigo Muramatsu, Osaka University
Assoc. Prof. Yasushi Makihara, Osaka University
Prof. Yasushi Yagi, Osaka University
Assoc. Prof. Sozo Inoue, Kyushu Institute of Technology
Tahera Hossain, Kyushu Institute of Technology
Anindya Das Antar, Dept. of EEE, University of Dhaka
Masud Ahmed, Dept. of EEE, University of Dhaka
Yuichi Hattori, Kyushu Institute of Technology
Recently, wearable computing resource, such as smartphone, is developing so quickly. People are using smartphone for communication, entertainment, business, travelling, browsing information. Although it has a huge benefit if we can use such wearable computing resource to support people life, the healthcare application is very limited. We would like to break the limitation and boost up the research to support human health. One of the important step for a healthcare system is to understand age and gender of the user who is wearing the sensor through gait. Gait is chosen since it is the most dominant daily activity, which is considered to contain not only identity but also physical, medical conditions.
The competition registration can be done by Google Form or email. If you would like to register by email, please send to wearablesensorchallengegmail.com with subject line as OU-ISIR-GAG, with the following information: (1) team name; (2) names, affiliations of all members; and (3) name, affiliation, email, phone number and mailing address of contact person. A reference number will be assigned to each team by our confirmation email.
The training datasets are publicly available, while test dataset will be uploaded. All these datasets were captured by the same sensors (IMUZ).
Participants can use both OU-ISIR inertial datasets:
(a) The OU-ISIR Gait Database, Inertial Sensor Dataset: http://www.am.sanken.osaka-u.ac.jp/BiometricDB/InertialGait.html
(b) The OU-ISIR Gait Database, Similar Action Inertial Dataset: http://www.am.sanken.osaka-u.ac.jp/BiometricDB/SimilarActionsInertialDB.html
A half of the dataset is similar to that of (a) and (b). The other half was captured in the wild on an almost flat ground; sensors were fixed in a backpack while their orientations will be unavailable. File format is the same as that of (a). The labels of this test dataset will not be released because we do not have release permission from most participants.
Download test dataset (expired)
We will evaluate the results separately for age and gender. The prediction results will be evaluated by mean absolute error for age and number of mistake for gender.
Participant can submit multiple solutions for age or gender. Submission of a technical paper (pdf, 2~4 pages) is required for each algorithm (age or gender) to explain the algorithm. Participant may decide not to disclose the method but need to briefly summarize the algorithm. A technical paper file should be named as reference number_X_Y, where X=AGE or GENDER and Y=1,2,.. the order of your algorithm. For examples, if you were given a reference number of GAG2019112701 and you want to submit two age solutions and one gender solution, you will submit 3 papers: GAG2019112701_AGE_1.pdf, GAG2019112701_AGE_2.pdf, and GAG2019112701_GENDER_1.pdf
Participant has to submit prediction result (age or gender) for all available data files of the test dataset in a prediction form. Number of prediction columns in the prediction form can be modified accordingly, depending on the number of solutions. The prediction form file should be named as reference number.csv
Prediction form and technical papers must be submitted by sending single or multiple emails to wearablesensorchallengegmail.com with the assigned reference number on the subject line.
Participants are encouraged to present their work at ICB 2019 conference orally onsite or online. Date and time will be announced.
Winner will be decided mainly by the prediction result, then partialy by the method details (novelty of algorithm, computation cost, … ), and presentation from presenters. A winner for age and a winner for gender will be selected.
|ID||Rank for Age||Rank for Gender||Team|
|GAG2019121202||1st||1st||Tim Van hamme (imec-DistriNet, KU Leuven), Giuseppe Garofalo (imec-DistriNet, KU Leuven), Davy Preuveneers (imec-DistriNet, KU Leuven), Enrique Argones Rúa (imec-COSIC, KU Leuven)|
|GAG2019120701||3rd||2nd||Takuya Yaguchi (So-net Media Networks Corp.), Yipeng Shen (University of Southampton), FangFei Liu (Taiping Financial Technology Service Co., Ltd.)|
|GAG2019121501||2nd||3rd||Sudeep Sarkar (Department of Computer Science, University of South Florida, USA), Ravichandran Subramanian (USF Alumnus)|