YOLOv7 Safety Use Case Tutorial

This repository contains a Jupyter Notebook that must be run alongside the SageMaker Image available at developer.sima.ai. The notebook walks through the following steps:

1. Retraining YOLOv7 on a safety-related dataset.
2. Exporting the trained model to ONNX.
3. Performing "Surgery" on the ONNX model (graph manipulation).
4. Quantizing the model.
5. Compiling the model to create a binary for the Sima.ai MLSoC platform.
6. Packaging and uploading the resulting .tar.gz artifact to an S3 bucket.

Prerequisites

• AWS account with permissions to use SageMaker.
• Access to the SageMaker Image from developer.sima.ai.
• An S3 bucket for uploading the final .tar.gz.

Usage

1. Launch SageMaker Notebook:

   • Select the Sima.ai image when creating a new SageMaker Notebook instance.

2. Clone this repository in your SageMaker environment:

   git clone https://github.com/<your-username>/<your-repo-name>.git
   cd <your-repo-name>