/YOLOv3-Custom-Object-Detection-with-ImageAI

Primary LanguageJupyter Notebook

🧠 YOLOv3 Custom Object Detection with ImageAI

This Jupyter Notebook demonstrates how to train and use a custom object detection model using ImageAI and YOLOv3. It guides you through dataset setup, training, and object detection on images.

📓 Contents

  • ImageAI.ipynb – Complete training and detection pipeline using YOLOv3
  • Built using the ImageAI library and a custom dataset
  • Outputs object predictions with bounding boxes and confidence scores

📦 Dataset Used

This notebook was originally tested using a custom dataset named hololens-yolo.zip, designed for detecting Microsoft HoloLens in images.

❗ Note: This dataset is not included

🛠️ Using Your Own Dataset

Yes! This notebook can be used to train on any custom object detection dataset, not just HoloLens.

To use your own dataset:

  1. Prepare your dataset in the following folder structure:

    your-dataset-name/
├── train/
├── test/
├── json/                  # Contains the detection config
└── models/                # Optionally include a pretrained YOLOv3 model

  2. Zip the folder (e.g., your-dataset-name.zip)

  3. In Google Colab, upload the zip file before running the notebook. You can do this by:

    • Clicking the 📁 file icon on the left sidebar
    • Clicking the 📤 upload button and selecting your .zip file

  4. Update the filename in the notebook:

    with zipfile.ZipFile('/content/your-dataset-name.zip', 'r') as zip_ref:
    zip_ref.extractall('/content/drive/MyDrive')