padelpose / README.md
josmazna's picture
Update README.md
1515783 verified
## Dataset Overview
This dataset contains sequences of MediaPipe pose landmarks for several padel strokes.
### Classes
- Forehand - filename: forehand (i).pkl
- Backhand - filename: backhand (i).pkl
- Forehand Volley - filename: fhvolley (i).pkl
- Backhand Volley - filename: bhvolley (i).pkl
- Bandeja - filename: bandeja (i).pkl
- Smash - filename: smash (i).pkl
### File Contents
Each `.pkl` file contains the `PoseLandmarkerResult` returned by `detect_for_video` from MediaPipe.
#### `PoseLandmarkerResult` Structure
- **Landmarks**:
- `Landmark #0`:
- `x`: 0.638852
- `y`: 0.671197
- `z`: 0.129959
- `visibility`: 0.9999997615814209
- `presence`: 0.9999984502792358
- `Landmark #1`:
- `x`: 0.634599
- `y`: 0.536441
- `z`: -0.06984
- `visibility`: 0.999909
- `presence`: 0.999958
- ... (33 landmarks per pose)
- **WorldLandmarks**:
- `Landmark #0`:
- `x`: 0.067485
- `y`: 0.031084
- `z`: 0.055223
- `visibility`: 0.9999997615814209
- `presence`: 0.9999984502792358
- `Landmark #1`:
- `x`: 0.063209
- `y`: -0.00382
- `z`: 0.020920
- `visibility`: 0.999976
- `presence`: 0.999998
- ... (33 world landmarks per pose)
More information on its format can be found [here](https://ai.google.dev/edge/mediapipe/solutions/vision/pose_landmarker/python).
## Loading Data
### Loading `PoseLandmarkerResult`
To load a `PoseLandmarkerResult` from a `.pkl` file, use the following code:
```python
import joblib
detection_results = joblib.load(pkl_filepath)
```
### Loading the Complete Dataset
To load the dataset into arrays `X`, `Y`, `Z`, and `labels`, use the sample code below. Make sure to point `folderpath` to the directory containing the dataset files.
```python
import os
import joblib
def build_coordinates_dataset(folderpath):
all_files = os.listdir(folderpath)
pkl_filepaths = [os.path.join(folderpath, pkl_filepath) for pkl_filepath in all_files if pkl_filepath.endswith(".pkl")]
print(f'{len(pkl_filepaths)} samples found')
X, Y, Z, labels = [], [], [], []
for pkl_filepath in pkl_filepaths:
detection_results = joblib.load(pkl_filepath)
try:
loaded_pose_landmarks = detection_results
x_frames, y_frames, z_frames = [], [], []
for frame_detection_result in loaded_pose_landmarks:
landmarks = frame_detection_result.pose_world_landmarks[0]
x_landmarks, y_landmarks, z_landmarks = [], [], []
for i in landmarks:
x_landmarks.append(i.x)
y_landmarks.append(i.y)
z_landmarks.append(i.z)
x_frames.append(x_landmarks)
y_frames.append(y_landmarks)
z_frames.append(z_landmarks)
X.append(x_frames)
Y.append(y_frames)
Z.append(z_frames)
labels.append(os.path.basename(pkl_filepath)[:3])
except Exception as e:
print(e)
print(f"Error loading {pkl_filepath}")
return X, Y, Z, labels
```
To load the dataset into arrays X, Y, Z, and labels, use the sample code below. Make sure to point folderpath to the directory containing the dataset files.