Awesome Open Source
Awesome Open Source

TrainYourOwnYOLO: Building a Custom Object Detector from Scratch License: CC BY 4.0

This repo let's you train a custom image detector using the state-of-the-art YOLOv3 computer vision algorithm. For a short write up check out this medium post. This repo works with TensorFlow 2.3 and Keras 2.4.

Before getting started:

  • 🍴 fork this repo so that you can use it as part of your own project.
  • ⭐ star this repo to get notifications on future improvements.

Pipeline Overview

To build and test your YOLO object detection algorithm follow the below steps:

  1. Image Annotation
    • Install Microsoft's Visual Object Tagging Tool (VoTT)
    • Annotate images
  2. Training
    • Download pre-trained weights
    • Train your custom YOLO model on annotated images
  3. Inference
    • Detect objects in new images and videos

Repo structure

  • 1_Image_Annotation: Scripts and instructions on annotating images
  • 2_Training: Scripts and instructions on training your YOLOv3 model
  • 3_Inference: Scripts and instructions on testing your trained YOLO model on new images and videos
  • Data: Input Data, Output Data, Model Weights and Results
  • Utils: Utility scripts used by main scripts

Getting Started

Google Colab Tutorial Open In Colab

With Google Colab you can skip most of the set up steps and start training your own model right away.

Requisites

The only hard requirement is a running version of python 3.6 or 3.7. To install python 3.7 go to

and follow the installation instructions. Note that this repo has only been tested with python 3.6 and python 3.7 thus it is recommened to use either python3.6 or python3.7.

To speed up training, it is recommended to use a GPU with CUDA support. For example on AWS you can use a p2.xlarge instance (Tesla K80 GPU with 12GB memory). Inference is very fast even on a CPU with approximately ~2 images per second. If you want to use your own machine, follow the instructions at tensorflow.org/install/gpu to install CUDA drivers. Make sure to install the correct version of CUDA and cuDNN.

Installation

Setting up Virtual Environment [Linux or Mac]

Clone this repo with:

git clone https://github.com/AntonMu/TrainYourOwnYOLO
cd TrainYourOwnYOLO/

Create Virtual (Linux/Mac) Environment:

python3 -m venv env
source env/bin/activate

Make sure that, from now on, you run all commands from within your virtual environment.

Setting up Virtual Environment [Windows]

Use the Github Desktop GUI to clone this repo to your local machine. Navigate to the TrainYourOwnYOLO project folder and open a power shell window by pressing Shift + Right Click and selecting Open PowerShell window here in the drop-down menu.

Create Virtual (Windows) Environment:

py -m venv env
.\env\Scripts\activate

PowerShell Make sure that, from now on, you run all commands from within your virtual environment.

Install Required Packages [Windows, Mac or Linux]

Install required packages (from within your virtual environment) via:

pip install -r requirements.txt

If this fails, you may have to upgrade your pip version first with pip install pip --upgrade.

Quick Start (Inference only)

To test the cat face detector on test images located in TrainYourOwnYOLO/Data/Source_Images/Test_Images run the Minimal_Example.py script in the root folder with:

python Minimal_Example.py

The outputs are saved in TrainYourOwnYOLO/Data/Source_Images/Test_Image_Detection_Results. This includes:

  • Cat pictures with bounding boxes around faces with confidence scores and
  • Detection_Results.csv file with file names and locations of bounding boxes.

If you want to detect cat faces in your own pictures, replace the cat images in Data/Source_Images/Test_Images with your own images.

Full Start (Training and Inference)

To train your own custom YOLO object detector please follow the instructions detailed in the three numbered subfolders of this repo:

To make everything run smoothly it is highly recommended to keep the original folder structure of this repo!

Each *.py script has various command line options that help tweak performance and change things such as input and output directories. All scripts are initialized with good default values that help accomplish all tasks as long as the original folder structure is preserved. To learn more about available command line options of a python script <script_name.py> run:

python <script_name.py> -h

NEW: Weights and Biases

TrainYourOwnYOLO supports Weights & Biases to track your experiments online. Sign up at wandb.ai to get an API key and run:

wandb -login <API_KEY>

where <API_KEY> is your Weights & Biases API key.

Multi-Stream-Multi-Model-Multi-GPU

If you want to run multiple streams in parallel, head over to github.com/bertelschmitt/multistreamYOLO. Thanks to @bertelschmitt for putting the work into this.

License

Unless explicitly stated otherwise at the top of a file, all code is licensed under CC BY 4.0. This repo makes use of ilmonteux/logohunter which itself is inspired by qqwweee/keras-yolo3.

Troubleshooting

  1. If you encounter any error, please make sure you follow the instructions exactly (word by word). Once you are familiar with the code, you're welcome to modify it as needed but in order to minimize error, I encourage you to not deviate from the instructions above. If you would like to file an issue, please use the provided template and make sure to fill out all fields.

  2. If you encounter a FileNotFoundError, Module not found or similar error, make sure that you did not change the folder structure. Your directory structure must look exactly like this:

    TrainYourOwnYOLO
    └─── 1_Image_Annotation
    └─── 2_Training
    └─── 3_Inference
    └─── Data
    └─── Utils
    

    If you use a different name such as e.g. TrainYourOwnYOLO-master you will have to specify the correct paths as command line arguments in every function call.

    Don't use spaces in file or folder names, i.e. instead of my folder use my_folder.

  3. If you are a Linux user and having trouble installing *.snap package files try:

    snap install --dangerous vott-2.1.0-linux.snap
    

    See Snap Tutorial for more information.

  4. If you have a newer version of python on your system, make sure that you create your virtual environment with version 3.7. You can use virtualenv for this:

    pip install virtualenv
    virtualenv env --python=python3.7
    

    Then follow the same steps as above.

Need more help? File an Issue!

If you would like to file an issue, please use the provided issue template and make sure to complete all fields. This makes it easier to reproduce the issue for someone trying to help you.

Issue

Issues without a completed issue template will be closed after 7 days.

Stay Up-to-Date

  • ⭐ star this repo to get notifications on future improvements and
  • 🍴 fork this repo if you like to use it as part of your own project.

CatVideo

Licensing

This work is licensed under a Creative Commons Attribution 4.0 International License. This means that you are free to:

  • Share — copy and redistribute the material in any medium or format
  • Adapt — remix, transform, and build upon the material for any purpose, even commercially.

Under the following terms:

  • Attribution

Cite as:

@misc{TrainYourOwnYOLO,
  title={TrainYourOwnYOLO: Building a Custom Object Detector from Scratch},
  author={Anton Muehlemann},
  year={2019},
  url={https://github.com/AntonMu/TrainYourOwnYOLO}
}

If your work doesn't include a citation list, simply link this github repo!

CC BY 4.0


Get A Weekly Email With Trending Projects For These Topics
No Spam. Unsubscribe easily at any time.
python (53,337) 
jupyter-notebook (6,166) 
deep-learning (3,908) 
keras (762) 
object-detection (475) 
gpu (357) 
yolo (125) 
transfer-learning (110) 
yolov3 (108) 
inference (97) 
tensorflow2 (94) 
deep-learning-tutorial (57) 
tf2 (39) 
detector (34) 
google-colab (29) 
keras-models (22)