Tehmas1012/action-recognition

Action Recognition & Caption Generation Model

Model Description

A multi-task deep learning model that performs:

1. Action Recognition - Classifies human actions in images
2. Image Captioning - Generates natural language descriptions

Architecture:

• Backbone: ResNet50 (pretrained on ImageNet)
• Action Head: Fully connected layers with dropout
• Caption Decoder: LSTM-based sequence generator

Intended Use

This model is designed for:

• Automated video content analysis
• Activity recognition in surveillance systems
• Image understanding and description generation
• Educational demonstrations of multi-task learning

Training Data

Provide information about the training data used.

Performance

Add performance metrics and evaluation results.

Usage

from transformers import pipeline

# Load the model
pipe = pipeline("task-name", model="your-username/model-name")

# Make predictions
result = pipe("Your input text here")
print(result)

Limitations

Describe any known limitations of the model.

Citation

If applicable, add citation information here.