Merge branch 'release/release-0.1'

dsmilab · May 19, 2018 · beac50d · beac50d
2 parents 4fce131 + 26a207a
commit beac50d
Show file tree

Hide file tree

Showing 55 changed files with 1,047,105 additions and 1 deletion.
diff --git a/.gitignore b/.gitignore
@@ -0,0 +1,10 @@
+
+\.idea/
+
+wearable-motion-sensors/
+
+data/
+
+ipynb/\.ipynb_checkpoints/
+
+ipynb/__pycache__/
diff --git a/README.md b/README.md
@@ -1 +1,54 @@
-# General-Movements
+# Taiko-Master
+
+This is a project teaching a newbie how to perform better in the video game [**Taiko no Tatsujin**](https://en.wikipedia.org/wiki/Taiko_no_Tatsujin).
+
+<br/>
+
+## Flowchart
+In general, follow the paper [Motion Primitive-Based Human Activity Recognition Using a Bag-of-Features Approach](https://dl.acm.org/citation.cfm?id=2110433).
+
+![](docs/flowchart.png)
+
+Belows are repos for collecting raw data from wearable devices.
+
+1. [beagle_bone_blue_data_acq](https://github.com/taoyilee/beagle_bone_blue_data_acq)
+
+	- [Introduction of beaglebone-blue](docs/144934_data.pdf): The official manual of BBB.
+
+	- [MPU9250A registration](docs/RM-MPU-9250A-00-v1.6.pdf): The register map and description of MPU-9250
+
+2. [USB-Video-Class-Capture](https://github.com/taoyilee/USB-Video-Class-Capture)
+
+![](docs/video_capture_sample.png)
+
+<br/>
+
+## Experiment
+Belows two animations are some extraced features with the specific entire play, and we plot vertical color lines to represent real true hit event. 
+
+![](docs/0420-axyz.gif)
+![](docs/0420-gxyz.gif)
+
+We can interpret the local event as the following figure.
+![](docs/time_series_sense.png)
+
+More observation can be checked at the following notebooks.
+
+1. [EDA](util/taiko_feature_eda.ipynb): show the box plot of engineering features.
+2. [BoF](util/taiko_BoF.ipynb): show the traning error about the bag-of-features model.
+3. [Confusion Matrix](util/kernel_test.ipynb): use model to predict other performance.
+
+In addtion, [Taiko-Time-Series-Analytics](https://github.com/taoyilee/Taiko-Time-Series-Analytics) is another related repo analyzing this data.
+
+<br/>
+
+## Result
+Use the last performance to train, we get
+
+![](docs/training_cnfm.png)
+
+However, predict a test performance, we get
+
+![](docs/test_cnfm.png)
+
+There is a long way to go! 
diff --git a/backup/bb_left_forearm_csv/data_2018_03_27_11_37_57.csv b/backup/bb_left_forearm_csv/data_2018_03_27_11_37_57.csv
diff --git a/backup/bb_left_forearm_csv/data_2018_03_27_13_25_43.csv b/backup/bb_left_forearm_csv/data_2018_03_27_13_25_43.csv
diff --git a/backup/bb_left_forearm_csv/data_2018_03_27_14_48_28.csv b/backup/bb_left_forearm_csv/data_2018_03_27_14_48_28.csv
diff --git a/backup/bb_left_forearm_csv/data_2018_03_27_16_12_02.csv b/backup/bb_left_forearm_csv/data_2018_03_27_16_12_02.csv
diff --git a/backup/bb_left_forearm_csv/data_2018_03_28_14_52_45.csv b/backup/bb_left_forearm_csv/data_2018_03_28_14_52_45.csv
diff --git a/backup/bb_left_forearm_csv/data_2018_03_28_15_52_48.csv b/backup/bb_left_forearm_csv/data_2018_03_28_15_52_48.csv
diff --git a/backup/bb_left_forearm_csv/data_2018_03_28_16_59_47.csv b/backup/bb_left_forearm_csv/data_2018_03_28_16_59_47.csv
diff --git a/backup/bb_left_forearm_csv/data_2018_03_31_15_29_36.csv b/backup/bb_left_forearm_csv/data_2018_03_31_15_29_36.csv
diff --git a/backup/bb_right_forearm_csv/data_2018_03_27_11_40_03.csv b/backup/bb_right_forearm_csv/data_2018_03_27_11_40_03.csv
diff --git a/backup/bb_right_forearm_csv/data_2018_03_27_13_20_28.csv b/backup/bb_right_forearm_csv/data_2018_03_27_13_20_28.csv
diff --git a/backup/bb_right_forearm_csv/data_2018_03_27_14_50_24.csv b/backup/bb_right_forearm_csv/data_2018_03_27_14_50_24.csv