two-wheelers_image-classifier Scones Unlimited

Project Overview

Scones Unlimited, a scone-delivery-focused logistics company, aims to optimize delivery operations using an image classification model. This model identifies delivery vehicles (bicycles and motorcycles) to route drivers efficiently, enhancing operational efficiency.

Technologies Used

• AWS Sagemaker: Building and deploying the image classification model
• AWS Lambda: Creating supporting services
• AWS Step Functions: Composing an event-driven application
• Python: Scripting and model development
• Jupyter Notebooks: Data exploration and model training

Project Steps

1. Data Staging

Prepare and upload the training and testing datasets to an S3 bucket. The datasets contain images of bicycles and motorcycles.

2. Model Training and Deployment

Use AWS Sagemaker to explore the data, train the image classification model, and deploy the trained model.

3. Lambdas and Step Function Workflow

Create AWS Lambda functions to preprocess input data and perform model inference. Define the Step Functions workflow to integrate these services.

4. Testing and Evaluation

Test the deployed model and evaluate its performance using a set of validation images.

5. Optional Challenge

Implement additional features or improvements to the model or workflow as per the optional challenge guidelines.

6. Cleanup Cloud Resources

Ensure all AWS resources created during the project are properly cleaned up to avoid unnecessary charges.

Getting Started

Prerequisites

• AWS Account
• Python 3.7+
• AWS CLI
• Git

Installation

1. Clone the repository:
   • git clone https://github.com/yourusername/scones-unlimited-image-classification.git
   • cd scones-unlimited-image-classification
2. Install dependencies:
   • pip install -r requirements.txt
3. Configure AWS CLI:
   • aws configure

Project Structure

• data/: Contains training and testing datasets
• notebooks/: Jupyter notebooks for data exploration and model training
• src/: Lambda functions and Step Function workflow definitions
• models/: Trained model files
• README.md: Project documentation
• requirements.txt: Project dependencies

Results

The image classification model successfully distinguishes between bicycles and motorcycles, aiding Scones Unlimited in optimizing delivery operations by routing drivers based on their vehicle type.

Contributing

Contributions are welcome! Please open an issue or submit a pull request for any changes or improvements.

License

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

Acknowledgements

• The team at Scones Unlimited
• AWS for providing the cloud infrastructure
• The open-source community for various tools and libraries