Evaluation Models Repository for YOLOv8

Overview

This repository is set up for evaluating YOLOv8 models on a specific dataset. It includes scripts for running predictions and computing evaluation scores.

Prerequisites

Getting Started

Step 1: Clone the Repository

Begin by cloning this repository to your local machine. Use the command:

git clone https://github.com/MateoLostanlen/eval_models.git
cd eval_models

Step 2: Install dependencies

pip install -e requirements.txt

Step 3: Download and Set Up Data

1. Download the Dataset
   Download the dataset DS-71c1fd51-v2 from Google Drive:

   https://drive.google.com/file/d/17syKwltw8Jv-nYlLjzxbXQDd4p9hmkKC/view?usp=sharing
   

   Unzip the contents into the Data directory within this repository.

2. Download Model Checkpoints
   Download the model checkpoints cp from Google Drive:

   https://drive.google.com/file/d/10zGvbR0nEvk5mPj7wEn3cHzWmyv4Hlqa/view?usp=sharing
   

   Unzip and place the folder in the same Data directory.

Your directory structure should now look like this:

repo-name/
│
└───Data/
    │
    ├───cp/
    │
    └───DS-71c1fd51-v2/

Step 4: Run Predictions

To compute predictions using the models and dataset, execute the following command:

python run_predictions.py

Step 5: Evaluate Results

After running predictions, evaluate the results using the Jupyter notebook provided:

jupyter notebook compute_scores.ipynb

Step 6: View erros using fiftyone

You can have a look to predictions erros using fiftyone running:

python fiftyone_create.py