日本データベース学会

dbjapanメーリングリストアーカイブ(2020年)

[dbjapan] Cooking challenge + ABC extended


DBJapanの皆様

行動とふるまいに関する国際会議ABC2020に向けて行われるクッキング行動認識チャレンジの案内をいたします。
モーションデータから、マルチラベルを認識する興味深いタスクとなっています。
賞金10万円と、モーションキャプチャ施設の利用権とそのための旅費が用意されています。

また、ABCの論文締め切りも1/31まで延長されました。
優秀な論文にはスポンサーから旅費も支援されます。
よろしくお願いいたします。

==============================================================
                                               CALL FOR PARTICIPATION
                           Cooking Activity Recognition Challenge
                                    https://abc-research.github.io/cook2020


                                   Jun. 22nd - 25th, 2020, Kitakyushu, Japan
             International Conference on Activity and Behavior Computing (ABC 2020)
                                   In conjunction with ICIEV&IVPR2020
==============================================================


Activity Recognition is the process of automatically inferring what a user is doing based on sensor observations. When your smartphone automatically identifies if you have been walking or running, it is doing activity recognition using the observations of its accelerometer sensor. Other applications of activity recognition include remote monitoring of daily activities for elders living alone and automatic record creation for nurses in hospitals. Although these applications only require the recognition of the activity being done, i.e.  cooking, such complex activities are usually made up of several smaller activities like ‘taking from the fridge’. Recognizing such steps can have several advantages.  For instance, in the scenario of an elder living alone recognizing the steps can be used to remind them of a missing step, or to ensure a healthy diet is being followed. In the scenario of the nursing record, recognizing the steps can be used for care quality assessment and for ensuring that safety protocols have been followed like washing the hands at the proper moments.

Current activity recognition systems focus on recognizing either the complex label (macro activity) or the small steps (micro activities) but their combined recognition is critical for analysis like the ones proposed. In fact, in a nursing scenario, washing the hands after taking blood is very different than doing it before, as it is mandatory. Therefore, in this challenge, we aim at the recognition of the macro and micro activities taking place during cooking sessions.


CHALLENGE
In this challenge, participants will create methods for recognizing both the macro and micro activities occurring in a 30 seconds window during a cooking session.
The evaluation will be done based on the average of the accuracy of the macro activity (classification) and that of the micro activities (multilabel classification)

PRIZES
The winner (one team representative) will get
- 100K yen, and
- be invited to our installations in Kitakyushu, Japan. The itinerary will be supported 250K yen. Read conditions in the web page.

PAPER FORMAT
- Challenge papers: 4-10 pages in a single column
- Please use Springer Contributed Book format
- Submit from Microsoft CMT for ICIEV2020 and select the ABC.

IMPORTANT DATES
Start of competition:                January 9, 2020
Registration closes:                  February 15
Test data sent to participants:  March 11, 2020
Submission of results:               March 18, 2020
Submission of paper:                March 25, 2020
Results of submission:              April 6, 2020
Camera-ready paper:                April 20, 2020
----
ORGANIZERS
Sozolab (sozolab.jp)
Larsen Team (https://team.inria.fr/larsen/)
==============================================================
-----
Sozo Inoue
九州工業大学 大学院生命体工学研究科 / 理化学研究所AIP / 情処UBI主査 井上創造