Skip to content

README.md

Latest commit

 

History

History
216 lines (151 loc) · 5.64 KB

README.md

File metadata and controls

216 lines (151 loc) · 5.64 KB

YOLO & Image Extractor Tool

📌 Description

The YOLO & Image Extractor Tool is a Windows batch script that combines multiple utilities into one program. It allows you to:

  • Extract images from video files using FFmpeg.
  • Train YOLO models with automated environment setup using Anaconda.
  • Install, uninstall, or check FFmpeg directly from the menu.

This tool is designed as a one-stop solution for dataset preparation, YOLO training, and environment setup.

✅ Requirements

  • Microsoft Windows 10/11
  • Anaconda (required for YOLO training)
  • FFmpeg (used for image extraction; can be installed via this tool)
  • Internet connection (for installing dependencies and downloading datasets)
  • GPU with CUDA support (recommended for YOLO training)

🎯 What It’s Used For

  • Extracting image datasets from video files.
  • Automatically setting up YOLO training environments with Python, PyTorch, and Ultralytics.
  • Downloading datasets directly from Roboflow.
  • Training YOLOv8 or YOLOv11 models with minimal manual setup.
  • Managing FFmpeg installation without leaving the tool.

🚀 How to Use

1. Main Menu

When you run the script, you’ll see:

================================================
                     MAIN MENU
================================================
[1] Extract images from video
[2] Run YOLO training
[3] Install/Uninstall FFmpeg
[4] Exit
================================================
Choose an option (1-4):

2. Option 1 – Extract Images from Video

  • Prompts for:

    • Video file path → e.g. C:\Videos\car.mp4
    • Output folder → e.g. "C:\Dataset\CarImages"
    • FPS (frames per second) → e.g. 2

  • Example review screen:

    Video file:    "C:\Videos\car.mp4"
Output folder: "C:\Dataset\CarImages"
FPS:           2
Base name:     car
Example file:  "C:\Dataset\CarImages\car_0001.png"

  • Example outcome in output folder:

    car_0001.png
car_0002.png
car_0003.png
...

Each image is a frame extracted at 2 FPS from the video.

3. Option 2 – Run YOLO Training

  • Prompts for:

    • YOLO version → enter 8 for YOLOv8 or 11 for YOLOv11.
    • Creates Conda environment (yolov8 or yolov11) if missing.
    • Installs dependencies: Ultralytics, PyTorch, Roboflow, Supervision.
    • Verifies CUDA and GPU availability.
    • Dataset → paste a Roboflow link, or press Enter to skip.
    • Epochs → e.g. 50.

  • Generates a train_yolo.py file with training, validation, and prediction steps.

  • Example training output:

    Epoch 1/50 - loss: 2.1
Epoch 25/50 - loss: 0.8
Epoch 50/50 - loss: 0.3
Training complete! Check the 'runs/detect' folder for results.

  • Example outcome:

    runs/detect/train/weights/best.pt    <- best trained model
runs/detect/train/results.png        <- training curves
runs/detect/predict/                 <- YOLO predictions on test images

4. Option 3 – Install/Uninstall FFmpeg

  • Menu:

    ================================================================
               INSTALL / UNINSTALL FFMPEG
================================================================
[1] Install FFmpeg
[2] Uninstall FFmpeg
[3] Check FFmpeg version
[4] Return to Main Menu
[5] Exit
================================================================

  • Install → downloads FFmpeg via Winget.

    Installing FFmpeg...
FFmpeg installed successfully.

  • Uninstall → removes FFmpeg if installed.

    Uninstalling FFmpeg...
Please restart.

  • Check version → shows installed FFmpeg details.

    ================================================================
ffmpeg version 8.0-full_build-www.gyan.dev
built with gcc 15.2.0 (Rev8, Built by MSYS2 project)
...
================================================================

5. Option 4 – Exit

  • Closes the tool.

Features

Image Extractor

  • Extract images from a video file at a specified FPS.
  • Automatically checks for FFmpeg and installs it if missing.
  • Allows specifying video file path, output folder, and FPS.

YOLO Runner

  • Detects Anaconda installation and prompts to install if missing.
  • Creates and activates a conda environment for YOLO training.
  • Installs necessary Python packages (ultralytics, supervision, roboflow, torch, torchvision).
  • Verifies CUDA availability.
  • Downloads dataset from Roboflow if a link is provided.
  • Allows setting the number of training epochs.
  • Generates a train_yolo.py script to run YOLO training, validation, and prediction.
  • Provides menu options to start training, return to YOLO runner, main menu, or exit.

FFmpeg Install/Uninstall

  • Installs or uninstalls FFmpeg using winget.
  • Checks if FFmpeg is installed before performing actions.
  • Allows checking the installed FFmpeg version.

Usage

Run the YOLOExtractor_Tool.cmd script. You will see a main menu with options to extract images, run YOLO training, manage FFmpeg, or exit.

Follow the prompts to enter required inputs and choose actions.

Requirements

  • Windows 10 or later with winget installed.
  • Anaconda installed for YOLO training.
  • Internet connection for downloading packages and datasets.

Notes

  • After installing or uninstalling FFmpeg, restart your command prompt or PowerShell window.
  • Ensure your video files and output folders exist and are accessible.
  • The script uses Python and conda commands; ensure they are in your PATH.

License

Please refer to the LICENSE file in the root BatchFile folder for license details.

Created by kerklangsi