YOLOv5 & YOLOv8 Object Detection

This repository demonstrates object detection using two versions of the YOLO model: YOLOv5 and YOLOv8 from Ultralytics. The scripts provided allow you to detect objects in images and videos using pre-trained models, and display or save the annotated outputs using OpenCV.

Table of Contents

• Installation
• Usage
• Detecting Objects in Images (YOLOv5)
• Processing Videos (YOLOv5 & YOLOv8)
• Dependencies
• Model Options
• YOLOv5 Models
• YOLOv8 Models
• Credits

Installation

1. Clone the repository: git clone https://github.com/shahinur-alam/YOLO.git
2. Install required dependencies: Make sure you have Python 3.x installed. Install the dependencies using pip: pip install -r requirements.txt
3. Install YOLOv5 and YOLOv8 dependencies:
   • For YOLOv5, the torch.hub feature is used to load the pre-trained model, so ensure PyTorch is installed:
   pip install torch torchvision torchaudio
   • For YOLOv8, the Ultralytics YOLO library is required:
   pip install ultralytics
   • Install OpenCV, which is required for video and image processing:
   pip install opencv-python numpy

Usage

Detecting Objects in Images (YOLOv5)

The code includes a function to detect objects in images using the YOLOv5 model.

1. Prepare the image: Save the image you want to detect objects in the root directory or specify its path in the script.
2. Uncomment and use the detect_objects function: Inside the code, the function detect_objects is currently commented out. Uncomment the relevant code and specify the path to the image you want to analyze.

   # Example usage
   image_path = "path/to/your/image.jpg"
   detect_objects(image_path)

3. Running the script: Once you’ve uncommented the function and provided the image path, run the script: python yolov5_image_detection.py
4. Output:

• The detected objects will be highlighted with bounding boxes.
• The output image is saved as output_image.jpg in the root directory.

Processing Videos (YOLOv5 & YOLOv8)

Both YOLOv5 and YOLOv8 models can be used to process video files. Below are examples for each model.

YOLOv5 Video Processing

1. Prepare the video file: Place the video you want to process in the project directory or reference its path in the script.
2. Set the video source: Modify the video_source variable in the script to point to your video file or use “0” for a live webcam feed.

   video_source = "path/to/your/video.mp4"

3. Running the script: Run the following command to start processing the video:

   python yolov5_video_detection.py

4. Output:

• The script will display each frame of the video with detected objects highlighted by bounding boxes.
• The processed video will be saved as output/output_video_v5.mp4.

YOLOv8 Video Processing

1. Specify the video path: Edit the video_path variable to point to your video file or webcam stream.

   video_path = "path/to/your/video.mp4"

2. Running the script: Run the following command to process the video with YOLOv8:

   python yolov8_video_detection.py

3. Output:

• The script will display the video frames with bounding boxes around detected objects.
• Press q to stop the video playback.

Dependencies

• Python 3.x
• PyTorch (for YOLOv5)
• Ultralytics YOLO library (for YOLOv8)
• OpenCV (opencv-python)
• NumPy

To install all dependencies:

pip install torch torchvision torchaudio ultralytics opencv-python numpy

Model Options

Both YOLOv5 and YOLOv8 support multiple model sizes, each with varying trade-offs in speed and accuracy.

YOLOv5 Models

In the YOLOv5 script, the following line loads the yolov5m.pt model:

model = torch.hub.load('ultralytics/yolov5', 'yolov5m', pretrained=True)

You can replace 'yolov5m' with other model sizes:

• 'yolov5n': Nano (fastest, least accurate)
• 'yolov5s': Small
• 'yolov5m': Medium
• 'yolov5l': Large
• 'yolov5x': Extra-large (slowest, most accurate)

YOLOv8 Models

For YOLOv8, the model is loaded with the following line:

model = YOLO('yolov8x.pt')  # 'n', 's', 'm', 'l', or 'x' for different sizes

You can replace 'yolov8x.pt' with the desired model size:

• yolov8n.pt: Nano (fastest, least accurate)
• yolov8s.pt: Small
• yolov8m.pt: Medium
• yolov8l.pt: Large
• yolov8x.pt: Extra-large (slowest, most accurate)

Credits

This repository utilizes YOLOv5 and YOLOv8 models, both developed by Ultralytics.

• YOLOv5 Official Repository
• YOLOv8 Official Repository

For more information on YOLO, please refer to the official YOLO documentation at Ultralytics.