identify_dance_styles

Model Card for Dance Style Classification Model

This model is a convolutional neural network (CNN) trained to classify dance styles from video frames. It predicts one of seven dance styles: ballet, breakdance, contemporary, hip hop, jazz, salsa, and tap.

Model Details

Model Description

The Dance Style Classification Model is designed to classify dance styles based on individual video frames. It uses a convolutional neural network (CNN) architecture optimized for multi-class classification tasks.

• Developed by: Caden Picard
• Funded by [optional]: Brightspeed Internal Project
• Shared by [optional]: Caden Picard
• Model type: Image Classification (Convolutional Neural Network)
• Language(s) (NLP): N/A
• License: CC BY-SA 4.0
• Finetuned from model [optional]: Custom-trained from scratch

Model Sources

• Repository: [More Information Needed]
• Paper [optional]: Not applicable
• Demo [optional]: Not available

Uses

Direct Use

This model can be used to predict dance styles from images of video frames. It is suitable for tasks like:

• Activity recognition in dance videos.
• Educational tools for teaching dance styles.
• Analyzing video content in multimedia databases.

Downstream Use [optional]

The model can be integrated into larger systems for real-time activity recognition or used as a component in video classification workflows.

Out-of-Scope Use

This model should not be used to identify individuals, infer personal attributes, or for any malicious or unethical purposes.

Bias, Risks, and Limitations

The dataset used for training may not fully represent the global diversity of dance styles. Biases could arise from the limited geographic or cultural sources of training data.

Recommendations

• Evaluate the model on additional datasets before deploying in production.
• Use the model responsibly, with consideration for the cultural nuances of dance styles.

How to Get Started with the Model

Use the following code snippet to get started with the model:

from tensorflow.keras.models import load_model
import cv2
import numpy as np

# Load the model
model = load_model("/path/to/your/model.h5")

# Preprocess an image
def preprocess_image(image_path, img_size=(128, 128)):
    img = cv2.imread(image_path)
    img = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)
    img = cv2.resize(img, img_size)
    img = img / 255.0
    return np.expand_dims(img, axis=0)

# Make a prediction
image_path = "/path/to/your/image.jpg"
input_image = preprocess_image(image_path)
predictions = model.predict(input_image)

# Decode predictions
class_names = ["ballet", "breakdance", "contemporary", "hip hop", "jazz", "salsa", "tap"]
predicted_class = class_names[np.argmax(predictions)]
print(f"Predicted class: {predicted_class}")

Training Details

Training Data

The model was trained on a dataset of video frames extracted from publicly available dance videos. Each frame is labeled with its corresponding dance style.

• Dataset Card: [https://huggingface.co/datasets/cadenlpicard/dance_images]

Training Procedure

The model was trained using a supervised learning approach with categorical cross-entropy loss and the Adam optimizer.

Training Hyperparameters

• Learning Rate: 0.001
• Batch Size: 32
• Epochs: 10

Testing Data, Factors & Metrics

Testing Data

The model was evaluated on a held-out test set comprising 20% of the dataset.

Factors

The evaluation was conducted across seven dance style categories.

Metrics

• Accuracy: Used to measure overall prediction accuracy.
• Classification Report: Includes precision, recall, and F1-score for each class.

Results

The model achieved a test accuracy of 87.5%.

Summary

The model performs well for most classes, with slightly lower performance for classes with fewer samples in the dataset.

Citation

BibTeX:

@model{caden_picard_dance_style_model,
  title={Dance Style Classification Model},
  author={Caden Picard},
  year={2024},
  howpublished={\url{https://huggingface.co/models/dance_style_model}}
}

APA:

Picard, C. (2024). Dance Style Classification Model. Retrieved from https://huggingface.co/models/dance_style_model