AWS Rekognition - Object and Face Detection

This repository contains Python scripts to use AWS Rekognition for detecting objects in images and faces in videos, displaying results with bounding boxes, and storing face detection results in a pandas DataFrame.

Table of Contents

• Introduction
• Prerequisites
• Installation
• Usage
  • Object Detection in Images
  • Face Detection in Videos
• License

Introduction

This repository provides two main functionalities using AWS Rekognition:

1. Object Detection in Images: Detects objects in an image and visualizes them with bounding boxes.
2. Face Detection in Videos: Detects faces in a video stored in an S3 bucket, retrieves detailed information about the faces, and stores the results in a pandas DataFrame.

Prerequisites

• AWS account with Rekognition and S3 services enabled.
• AWS credentials.
• Python 3.x installed.
• Required Python packages:
  • boto3
  • pandas
  • matplotlib
  • Pillow

Installation

1. Clone the repository:

   git clone https://github.com/yourusername/aws-rekognition-detection.git cd aws-rekognition-detection

2. Install the required packages:

   pip install boto3 pandas matplotlib Pillow

Usage

Object Detection in Images

1. Update the script detect_objects.py with the absolute paths of your AWS credentials CSV file and the image file:

   credentials_path = r'path/to/your/credentials.csv' photo_path = r'path/to/your/photo.jpg'

2. Run the script:

   python detect_objects.py

3. The script will display the image with bounding boxes around the detected objects.

Face Detection in Videos

1. Update the script detect_faces.py with your AWS credentials, region name, S3 bucket name, and video file name:

   access_key = 'your-aws-access-key'
secret_key = 'your-aws-secret-key' region_name = 'your-aws-region-name' bucket_name = 'your-s3-bucket-name' video_file = 'your-video-file.mp4'

2. Run the script:

   python detect_faces.py

3. The script will print the first row of the DataFrame containing the detected faces' information.

License

This project is licensed under the MIT License - see the LICENSE file for details