This is the main repository for CG3002 project at the National University of Singapore (AY2018/2019 Semester 1). CG3002 is the capstone module for the Computer Engineering programme.
This project detects labelled dance moves performed by user.
Accomplished with the use of:
- Arduino: Obtain values from sensor and output to Raspberry Pi
- Raspberry Pi: Load sensor values to Machine Learning model, send prediction to server
- Machine Learning: Process raw data from sensor and output prediction
For more instructions on the software component, please see here.
- Make sure you have installed the relevant dependencies for each component of this system;
- First, put all data under
software/data/raw
;- You can download the dataset as a zip file from here. Unzip and put it under
software/data/raw
.
- You can download the dataset as a zip file from here. Unzip and put it under
- Then, run
software/extract_raw.py
to extract features intosoftware/data/extract
;- Alternatively, you can download the extracted data as a zip file from here. Unzip and put it under
software/data/extract
.
- Alternatively, you can download the extracted data as a zip file from here. Unzip and put it under
- After that, run
software/train.py
to train the model (which will be stored under thesoftware/model
folder);- Alternatively, you can download the pre-trained model from here. Put it under
software/model
.
- Alternatively, you can download the pre-trained model from here. Put it under
- During demo, do the following:
- Run
ArduinoRpiCommunications
to start sending data from Arduino to the Raspberry Pi; - Run the following two programs at the same time:
python software/main.py
python comms/server-client/final_eval_server.py [IP address] 3002 09
- Run
Name | Sub-team |
---|---|
Leow Zheng Yu | Hardware |
Xiang Hailin | Hardware |
Gauri Joshi | Software |
Niu Yunpeng | Software |
Chua Kun Hong | Communications |
Ang Zhi Yuan | Communications |