In the age of social media, the demand for high-quality content with engaging captions is paramount for audience engagement and community building. This research explores the transformative impact of Generative AI, exceeding initial expectations.
The study focuses on CNN and RNN, specifically LSTM, for image detection and caption generation.
The model’s central purpose is to detect objects in images, predict their relationships, and produce meaningful captions. The novelty lies in enhancing these generated captions through the utilization of an open-source API key, connecting to an advanced Large Language Model (LLM) like Chat GPT-4.
This method not only enhances engagement by generating captions that are relevant, informative, engaging, and ensures the generation of ethical content by incorporating safeguards against bias, misinformation, and harm.
- Image detection and object recognition using CNN.
- Caption generation using LSTM.
- Integration with Chat GPT-4 for enhancing generated captions.
- Safeguards against bias, misinformation, and harm.
-
Clone this repository to your local machine:
-
Install the required dependencies:
-
Obtain necessary API keys for Chat GPT-4 and any other services used in the project.
-
Replace placeholders in the code with your actual API keys and configurations.
- Run the provided scripts or notebooks to train the model, detect objects in images, generate captions, and integrate with Chat GPT-3.5.
- Follow the instructions in the code comments or README files for each component of the project.
Research Paper: Journal of Research Administration
Citation:
@article{kabil2024generative,
author = {Kabil C.A and Vairachilai S},
title = {Exploring Generative AI Approaches for Crafting Image Captions: A Research Inquiry into Ethical, Engaging Content Generation},
journal = {Journal of Research Administration},
volume = {6},
pages = {5545--5552},
year = {2024}
}