2022小美赛C题Classify Human Activities对人类活动进行分类思路分享
点击链接【2022小美赛数学建模思路分享】：https://jq.qq.com/?_wv=1027&k=pYYvA9gJ
点击链接【2022小美赛数学建模思路分享】：https://jq.qq.com/?_wv=1027&k=pYYvA9gJ
点击链接【2022小美赛数学建模思思路分享】：https://jq.qq.com/?_wv=1027&k=pYYvA9gJ
One important aspect of human behavior understanding is the recognition and
monitoring of daily activities. A wearable activity recognition system can im
prove the quality of life in many critical areas, such as ambulatory monitor
ing, home-based rehabilitation, and fall detection. Inertial sensor based activ
ity recognition systems are used in monitoring and observation of the elderly
remotely by personal alarm systems[1], detection and classifification of falls[2],
medical diagnosis and treatment[3], monitoring children remotely at home or in
school, rehabilitation and physical therapy , biomechanics research, ergonomics,
sports science, ballet and dance, animation, fifilm making, TV, live entertain
ment, virtual reality, and computer games[4]. We try to use miniature inertial
sensors and magnetometers positioned on difffferent parts of the body to classify
human activities, the following data were obtained.
Each of the 19 activities is performed by eight subjects (4 female, 4 male,
between the ages 20 and 30) for 5 minutes. Total signal duration is 5 minutes
for each activity of each subject. The subjects are asked to perform the activ
ities in their own style and were not restricted on how the activities should be
performed. For this reason, there are inter-subject variations in the speeds and
amplitudes of some activities.
Sensor units are calibrated to acquire data at 25 Hz sampling frequency.
The 5-min signals are divided into 5-sec segments so that 480(= 60 × 8) signal
segments are obtained for each activity.
The 19 activities are:
1. Sitting (A1);
2. Standing (A2);
3. Lying on back (A3);
4. Lying on right side (A4);
5. Ascending stairs (A5);
16. Descending stairs (A6);
7. Standing in an elevator still (A7);
8. Moving around in an elevator (A8);
9. Walking in a parking lot (A9);
10. Walking on a treadmill with a speed of 4 km/h in flflat position and 15 deg
inclined positions (A10);
11. Walking on a treadmill with a speed of 4 km/h in 15 deg inclined positions
(A11);
12. Running on a treadmill with a speed of 8 km/h (A12);
13. Exercising on a stepper (A13);
14. Exercising on a cross trainer (A14);
15. Cycling on an exercise bike in horizontal position (A15);
16. Cycling on an exercise bike in vertical position (A16);
17. Rowing (A17);
18. Jumping (A18);
19. Playing basketball (A19).
Your team are asked to develop a reasonable mathematical model to solve
the following problems.
1. Please design a set of features and an effiffifficient algorithm in order to classify
the 19 types of human actions from the data of these body-worn sensors.
2. Because of the high cost of the data, we need to make the model have
a good generalization ability with a limited data set. We need to study
and evaluate this problem specififically. Please design a feasible method to
evaluate the generalization ability of your model.
3. Please study and overcome the overfifitting problem so that your classififi-
cation algorithm can be widely used on the problem of people’s action
classifification.
The complete data can be downloaded through the following link:
https://caiyun.139.com/m/i?0F5CJUOrpy8oq
人类行为理解的一个重要方面是对日常活动的识别和监控。一个可穿戴的活动识别
系统可以提高许多关键领域的生活质量，如动态监测、家庭康复和跌倒检测。基于
惯性传感器的活动识别系统用于监测和观察老年人远程个人报警系统[1]，检测和
分类瀑布[2]，医疗诊断和治疗[3]，监控儿童远程在家里或在学校，康复和物理治
疗，生物力学研究、人体工程学、体育科学、芭蕾和舞蹈、动画、电影制作、电视
、现场娱乐、虚拟现实和电脑游戏[4]。我们尝试使用定位于人体不同部位的微型
惯性传感器和磁力计来对人类活动进行分类，从而获得了以下数据。
每项19项活动分别由8名受试者（
4名女性，4名男性，年龄在20-30岁之间）进
行，持续5分钟。每个受试者每次活动的总信号持续时间为5分钟。受试者被要求以
自己的风格进行活动，而不受如何进行活动的限制。因此，一些活动的速度和振幅
存在着学科间的变化。
传感器单元被校准，以获取25 Hz采样频率的数据。5分钟的信号被分成5秒的片
段，以便为每个活动获得480个（
=60×8）的信号片段。
这19项活动包括：
1.坐（
A1）；
2.站（
A2）；
3.躺在背上（
A3）
4.右侧侧卧（
A4）；
5.上升楼梯（
A5）；2
6.下楼梯（
A6）；
7.站在电梯里（
A7）；
8.在电梯中四处移动（
A8）；
9.在停车场内行走（
A9）；
1 0 . 在跑步机上平速4 km/h，倾斜15度（
A10）；
11.在跑步机上以4 km/h的速度行走，倾斜15度（
A11）；
12.在跑步机上跑步，速度为8 km/h（
A12）；
13.练习使用一个步进器（
A13）；
14.使用交叉训练器进行练习（
A14）；
15.在水平位置骑运动自行车（
A15）；
16.以垂直位置骑运动自行车（
A16）；
17.划船（
A17）；
18.跳跃（
A18）；
19.打篮球（
A19）。
您的团队被要求开发一个合理的数学模型来解决以下问题。
1.请设计一套特征和一个有效的算法，以便从这些磨损传感器的数据中分类19
种类型的人体行为。
2.由于数据的高成本，我们需要使模型在有限的数据集下具有良好的泛化能力
。我们需要具体地研究和评估这个问题。请设计一种可行的方法来评估您的
模型的泛化能力。
3 . 请研究并克服过拟合问题，使您的分类算法能够广泛应用于人的动作分类问
题
