nuggy875/motion-classification

Real-Time Classification System for Motion Sensor Data

Proprio Motion Classification

Proprio Motion Classification sets up a server to classify streaming motion data into meaningful metrics. The inputs are a session id and user id. The outputs are a JSON with the associated metrics. Currently, Proprio Motion Classification is designed for transforming accelerometer and gyroscope to tennis strokes. The output will consist of information on the time and type of each stroke, rally and game information, and user and watch hardware information.

Set up environment on Ubuntu 14.04

./setup.sh

Set up and run classification server

pip install -r requirements
cd motion-classification
cp config.py.template config.py

Gunicorn

gunicorn --bind 0.0.0.0:5000 -t 6000 -w 4 --log-level="debug" main:app randomforest

Flask (deprecated)

python main.py randomforest

Finally, make sure to edit the config.py for the local dev, server dev, or server production as outlined below:

Development on Local Computer

Change the config.py file such that

es_url = 'localhost'
es_index = 'proprio'
analysis_port = 5000

Development on Server

Change the config.py file such that

es_url = 'IP.Address.of.Data.Server'
es_index = 'proprio_dev'
analysis_port = 6000

Development on Production

Change the config.py file such that

es_url = 'IP.Address.of.Data.Server'
es_index = 'proprio'
analysis_port = 5000

API Details

• URL

  /tennnis

• Method:

  POST

• Data Params

  {'session':session_id_from_proprio_app}

• Success Response:

  • Code: 200
    Content: {'rating': '4.5', 'userId': '10553...', 'session': '1463957174937', 'bezel': 'Towards Wrist', 'heightInches': '5\' 9"', 'privacy': 'No', 'samples': 321, 'total_points': 3, 'timestamped': {'Backhands': [], 'Forehands': [{'max_speed': 74.1, 'max_acceleration': 110.3, 'time': 1463957178935}, {'max_speed': 134.7, 'max_acceleration': 230.8, 'time': 1463957181836}], 'Serves': [{'max_speed': 148, 'max_acceleration': 276.6, 'time': 1463957180618}]}, 'product': 'bowfin', 'max_time': 1463957186225, 'activity': 'Tennis', 'hand': 'Lefty', 'aggregate': {'Backhands': 0, 'Forehands': 2, 'Serves': 1}, 'manufacturer': 'Motorola', 'userName': 'Superman', 'mean_rally': 1.3, 'gender': 'Male', 'age': '25', 'calories': 21.6, 'max_rally': 2, 'rallies': [{'game_number': 1, 'game_type': 'Return', 'rally': [{'ballspeed': 74.1, 'stroke': 'forehand', 'time': 1463957178935}]}, {'game_number': 2, 'game_type': 'Service', 'rally': [{'ballspeed': 148.0, 'stroke': 'serve', 'time': 1463957180618}]}, {'game_number': 2, 'game_type': 'Service', 'rally': [{'ballspeed': 134.7, 'stroke': 'forehand', 'time': 1463957181836}]}], 'model': 'Moto 360'}

• Error Response:

  • Code: 404 NOT FOUND
    Content: { error : "Session doesn't exist" }

  OR

  • Code: 401 UNAUTHORIZED
    Content: { error : "You are unauthorized to make this request." }

• Sample Call:

    $.ajax({
      url: "/tennis",
      dataType: "json",
      type : "POST",
      success : function(r) {
        console.log(r);
      }
    });

Classifier

python classifier.py name_of_new_classifier plot