Flode-Labs / auto-labeler

Label your images using GPT-4!

Auto Labeler

Auto Labeler is an automated image annotation tool that leverages the power of OpenAI's GPT-4 Vision API to detect objects in images and provide bounding box annotations.

Features

• Object Detection: Automatically identifies objects in images.
• Bounding Box Annotations: Generates bounding boxes around detected objects.
• Multiple Formats: Supports saving annotations in both TensorFlow Object Detection and JSON formats.
• Custom Labels: Allows for specifying custom object labels for detection.

Getting Started

Prerequisites

• Python 3.6 or higher

Installation

Clone the repository to your local machine:

git clone https://github.com/Flode-Labs/auto-labeler.git

Usage

1. Install required packages:

pip install -r requirements.txt

2. Set your OpenAI API key:

api_key = "YOUR_API_KEY"

3. Specify the path of the folder with the images that you want to label:

input_folder = '/path/to/input/folder'

4. Specify the path of the folder where you want to save the labels:

output_folder = '/path/to/output/folder'

5. Specify the format in which you want to save the labels:

output_format = 'json' # or 'tf'

6. Define the list of labels you want to detect:

labels = ['cat', 'dog', 'bird']

7. Run the main script:

python main.py

Functions

• encode_image(image_path): Encodes the image to base64 format.
• detect_objects(image_path, labels): Detects objects in the image as specified by labels.
• save_tf_annotations(...): Saves annotations in TensorFlow Object Detection XML format.
• save_json_annotations(...): Saves annotations in JSON format.
• process_images_in_folder(...): Processes multiple images in a specified folder.

Acknowledgments

• OpenAI for providing the GPT-4 Vision API.
• The Python community for maintaining and improving the libraries used in this project.

---

Ensure you replace YOUR_API_KEY with your actual OpenAI API key.