The 9th OSS(Open Source Software) Grand Developers Challenge μΌμ±μ μ κΈ°μ μ μκ³Όμ λ³Έμ μ§μΆ νλ‘μ νΈ
ML(Classifier) + Web(Backend, Server) + Smartwatch App(Frontend, Client)
- π This repo is part of the full project code.
λ΄μ© | URL |
---|---|
1. λνμκ° | https://www.oss.kr/notice/show/6008d9bc-66f0-4373-a9df-19a8973c7038 |
2. μμ°μμ | https://youtu.be/p5vPWqi1B6w |
3. λ°νμλ£ | https://www.slideshare.net/SuHyunCho2/sws-56703648#1 |
4. κ°λ°λ¬Έμ | https://www.slideshare.net/secret/bsfNKp1uR5Y1q8 |
5. λν μ΄ν λ Όλ¬Έμμ±νμ¬ κΈ°λ‘ | https://www.slideshare.net/SuHyunCho2/recognition-of-anaerobic-based-on-machine-learning-using-smart-watch-sensor-data ([paper site1] / [paper site2]) |
AI personal trainer App based on Machine Learning using Samsung tizen smart watch.
μΌμ± νμ΄μ OS
κΈ°λ°μ Gear(Gear2 , Gear S, Gear S2) μ ν리μΌμ΄μ tizen-sdk-2.3.1
- Device Optimization completed on
samsung smartwatch
version(Samsung Galaxy Gear 2, Gear S, Gear S2) - Tizen app type: Companion(Operating with Samsung Galaxy S4(android 4.4))
λ¨Έμ λ¬λ κΈ°λ° μΈκ³΅μ§λ₯ νΌνΈλμ€ ν¬μ€ μ½μΉ μ΄ν리μΌμ΄μ
.
3μΆκ°μλμΌμ 3μΆμμ΄λ‘μ€μ½νμΌμλ₯Ό κΈ°λ°μΌλ‘ μ¬μ©μμ λͺ¨μ
μ μ€μκ° νΈλνΉ λ° μ€μΌμ₯΄λ§νμ¬ ‘무μ¨μ΄λ’μ ‘λͺν’νλμ§ κ·Έλ¦¬κ³ ‘μΉΌλ‘리 μλͺ¨λ’κΉμ§ μ€μ€λ‘ μμμ νλ¨νκ³ κΈ°λ‘νμ¬ κ΄λ¦¬.
λν μ¬μ©μκ° μ΄λμ μμν ν μ¬λ°μλ₯Ό μμνμ¬ μ΄λμ μ΄μ§ν μμλλ‘ μ¬μ©μμ μμλ νκ· μ¬μ₯λ°λμμ κ°μ₯ λΉμ·ν BPMμ ν΄λΉλλ μμ
μ μ€μ€λ‘ μ°Ύκ³ μμμ μ¬μ.
- Performance(Accuracy): about 96.7% for unseen data [2016. 10]
- Model type:
Discriminative Model
( P ( y | X ) ) for inference - Learning Type: Classification on Supervised Learning.
- Using Dimension Reduction Skills e.g. PCA, LDA(Fisher’s LDA)
- Using Kernel Tricks e.g. linear and rbf
- Hybrid Stacking Model based on SVM(Support Vector Machine) Framework and others
- client side
- python 3.4 / 2.7
- tizen 2.3.1
- java 8
- android 4.4
- windows 7
- server side
- ubuntu 14
- AWS EC2 free tier
- flask 0.9
- nginx 1.4.6
- mariadb 5.5.44
- uwsgi 1.9.17.1
- sqlalchemy 0.15
- Samsung Developers
- Samsung Tizen Forum
- book 1 (kor)
- book 2 (kor)
- book 3 (kor)
Leave a Reply