NTNU EfficientDet training

Introduction

Deep learning project about animal behaviour.

Features

For now, the project includes:

• EfficientDet model (no backbone) training and saving
• EfficientDet model (no backbone) evaluation (including metrics and image testing)
• EfficientNet backbone training and saving
• EfficientDet with EfficientNet backbone training and saving
• EfficientDet with EfficientNet backbone evaluation (including metrics and image testing)

Prerequisites

Before you begin, ensure you have met the following requirement:

• Python 3.10+

Installation

Step-by-step guide on how to install the project:

git clone https://github.com/antoinedenovembre/NTNU_project.git
cd NTNU_project
python install_requirements.py

/!\ Important: You also need to make sure you have the following structure, including the data folder

NTNU_project_light
│   README.md
│   main.py
│   install_requirements.py
│
└───data
│   └───backbone
│   │   │   image1.jpg/png/...
│   │   │   ...
│   │
│   └───effdet
│       └─── train
│       │    └─── annotations
│       │    │    │   annotations.json
│       │    │    │   ...
│       │    │
│       │    └─── images
│       │         │   image1.jpg/png/...
│       │         │   ...
│       │
│       └─── test
│            └─── annotations
│            │    │   annotations.json
│            │    │   ...
│            │
│            └─── images
│                 │   image1.jpg/png/...
│                 │   ...
│
└───efficient_det
│
└───barlow
│
└───utils
│
└───scripts
│
└───documentation
│
└───output

The annotations shall have the following structure

{
  "annotations": [
    {
      "area": 87292,
      "bbox": [
        576,
        98,
        547,
        204
      ],
      "category_id": 6,
      "id": 1, <!-- Should be the number of the annotation -->
      "image_id": 2, <!-- Should be the number of the image -->
      "iscrowd": 0
    },
    ...
    ]
}

Usage

Here is the command to run the project:

python main.py

Documentation

Overall view of the project: Documentation - Overall view

Technical documentation about evaluation metrics and loss function: Documentation - Metrics

Q&As

What PC can be used?

• Ideally one with a sufficient GPU, like NVIDIA RTX 2080Ti

What OS can be used?

• Any Linux distro should do the trick, but I recommend using Ubuntu 22.04+

Contributing

This repository is a fork from this repository

Authors and Acknowledgment

Show your appreciation to those who have contributed to the project.

• @FayazRahman - Contribution
• Dev Narayan Chakkalakkal Binu - Contribution
• @antoinedenovembre - Contribution