Datasets:
Modalities:
Text
Formats:
text
Size:
< 1K
Tags:
humanoid-robotics
fall-prediction
machine-learning
sensor-data
robotics
temporal-convolutional-networks
License:
OliverUrbann
commited on
Commit
•
bc5a291
1
Parent(s):
2f6a432
doc and eaxmple
Browse files- README.md +51 -0
- requirements.txt +3 -0
- usage_example.py +54 -0
README.md
CHANGED
@@ -38,6 +38,57 @@ This dataset is shared under the **apache-2.0** license, allowing use and modifi
|
|
38 |
## Citation
|
39 |
If you use this dataset in your research, please cite it as follows:
|
40 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
41 |
---
|
42 |
license: apache-2.0
|
43 |
---
|
|
|
38 |
## Citation
|
39 |
If you use this dataset in your research, please cite it as follows:
|
40 |
|
41 |
+
|
42 |
+
## How to Use the Dataset
|
43 |
+
|
44 |
+
To get started with the **Fall Prediction Dataset for Humanoid Robots**, follow the steps below:
|
45 |
+
|
46 |
+
### 1. Set Up a Virtual Environment
|
47 |
+
|
48 |
+
It's recommended to create a virtual environment to isolate dependencies. You can do this with the following command:
|
49 |
+
|
50 |
+
```bash
|
51 |
+
python -m venv .venv
|
52 |
+
```
|
53 |
+
|
54 |
+
After creating the virtual environment, activate it:
|
55 |
+
|
56 |
+
- On **Windows**:
|
57 |
+
```bash
|
58 |
+
.venv\Scripts\activate
|
59 |
+
```
|
60 |
+
|
61 |
+
- On **macOS/Linux**:
|
62 |
+
```bash
|
63 |
+
source .venv/bin/activate
|
64 |
+
```
|
65 |
+
|
66 |
+
### 2. Install Dependencies
|
67 |
+
|
68 |
+
Once the virtual environment is active, install the necessary packages by running:
|
69 |
+
|
70 |
+
```bash
|
71 |
+
pip install -r requirements.txt
|
72 |
+
```
|
73 |
+
|
74 |
+
### 3. Run the Example Script
|
75 |
+
|
76 |
+
To load and use the dataset for training a simple LSTM model, run the `usage_example.py` script:
|
77 |
+
|
78 |
+
```bash
|
79 |
+
python usage_example.py
|
80 |
+
```
|
81 |
+
|
82 |
+
This script demonstrates how to:
|
83 |
+
- Load the dataset
|
84 |
+
- Select the relevant sensor columns
|
85 |
+
- Split the data into training and test sets
|
86 |
+
- Train a basic LSTM model to predict falls
|
87 |
+
- Evaluate the model on the test set
|
88 |
+
|
89 |
+
Make sure to check the script and adjust the dataset paths if necessary. For further details, see the comments within the script.
|
90 |
+
|
91 |
+
|
92 |
---
|
93 |
license: apache-2.0
|
94 |
---
|
requirements.txt
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
pandas
|
2 |
+
tensorflow
|
3 |
+
scikit-learn
|
usage_example.py
ADDED
@@ -0,0 +1,54 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
# Usage Example for the Fall Prediction Dataset
|
2 |
+
# Please install dependencies before:
|
3 |
+
# pip install -r requirements.txt
|
4 |
+
|
5 |
+
# Import necessary libraries
|
6 |
+
import pandas as pd
|
7 |
+
import tensorflow as tf
|
8 |
+
from tensorflow.keras.models import Sequential
|
9 |
+
from tensorflow.keras.layers import Dense, LSTM, Dropout
|
10 |
+
from sklearn.model_selection import train_test_split
|
11 |
+
|
12 |
+
# Load the dataset from Huggingface or a local file path
|
13 |
+
# Example for local loading; replace with Huggingface dataset call if applicable
|
14 |
+
real_data = pd.read_csv('dataset.csv.bz2', compression='bz2')
|
15 |
+
|
16 |
+
# Preview the dataset
|
17 |
+
print(real_data.head())
|
18 |
+
|
19 |
+
# Select relevant columns (replace these with actual column names from your dataset)
|
20 |
+
# Here we assume that the dataset contains sensor readings like gyroscope and accelerometer data
|
21 |
+
relevant_columns = ['gyro_x', 'gyro_y', 'gyro_z', 'acc_x', 'acc_y', 'acc_z', 'upright']
|
22 |
+
sensordata = real_data[relevant_columns]
|
23 |
+
|
24 |
+
# Split the data into features (X) and labels (y)
|
25 |
+
# 'fall_label' is assumed to be the column indicating whether a fall occurred
|
26 |
+
X = sensordata.drop(columns=['upright']) # Replace 'fall_label' with the actual label column
|
27 |
+
y = sensordata['upright']
|
28 |
+
|
29 |
+
# Split data into training and test sets
|
30 |
+
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42)
|
31 |
+
|
32 |
+
# Reshape data for LSTM input (assuming time-series data)
|
33 |
+
# Adjust the reshaping based on your dataset structure
|
34 |
+
X_train = X_train.values.reshape(X_train.shape[0], X_train.shape[1], 1)
|
35 |
+
X_test = X_test.values.reshape(X_test.shape[0], X_test.shape[1], 1)
|
36 |
+
|
37 |
+
# Define a simple LSTM model
|
38 |
+
model = Sequential()
|
39 |
+
model.add(LSTM(64, input_shape=(X_train.shape[1], 1)))
|
40 |
+
model.add(Dropout(0.2))
|
41 |
+
model.add(Dense(1, activation='sigmoid'))
|
42 |
+
|
43 |
+
# Compile the model
|
44 |
+
model.compile(optimizer='adam', loss='binary_crossentropy', metrics=['accuracy'])
|
45 |
+
|
46 |
+
# Train the model
|
47 |
+
history = model.fit(X_train, y_train, epochs=10, batch_size=64, validation_data=(X_test, y_test))
|
48 |
+
|
49 |
+
# Evaluate the model on the test set
|
50 |
+
loss, accuracy = model.evaluate(X_test, y_test)
|
51 |
+
print(f"Test Accuracy: {accuracy * 100:.2f}%")
|
52 |
+
|
53 |
+
# You can save the model if needed
|
54 |
+
# model.save('fall_prediction_model.h5')
|