YOLO Slicing - Segmentation Scripts

This repository contains three Python scripts utilizing YOLOv11s-seg for object segmentation in videos. These scripts progressively enhance object detection accuracy through slicing techniques, making them effective for detecting small or partially visible objects.

<img src="https://raw.githubusercontent.com/jsammarco/YoloSlicing/57d5a2bfbf7e0ba5aa967d111868a0d3c575fbdd/banner.gif">

Video Explanation

---

For a detailed explanation and demonstration of these scripts, watch the YouTube video.

Scripts Overview

1. yoloDetect.py

• Description: Performs standard YOLO segmentation on an entire video frame. Ideal for detecting objects in scenes with minimal occlusion or overlap.
• Key Features:
  • Processes the entire frame in one step.
  • Outputs annotated videos with detected objects and confidence scores.
  • Simple and efficient for general use cases.
• Usage: Best suited for quick segmentation tasks with standard video footage.

---

2. yoloSliceDetect.py

• Description: Enhances object detection by dividing the frame into 4 overlapping slices (2 rows × 2 columns). This approach helps in detecting smaller objects.
• Key Features:
  • Utilizes 10% overlap between slices for seamless object detection.
  • Processes each slice independently to improve accuracy for small objects.
  • Combines results from all slices into a unified annotated video.
• Usage: Recommended for videos with smaller or partially obscured objects.

---

3. yoloSuperSliceDetect.py

• Description: Maximizes detection accuracy by dividing the frame into 12 overlapping slices (4 rows × 3 columns). This method is optimized for complex scenes with very small or occluded objects.
• Key Features:
  • Uses 10% overlap between slices for superior coverage.
  • Processes a higher number of slices to ensure no small object is missed.
  • Balances detection quality, speed, and quantity effectively.
• Usage: Ideal for videos with high object density or complex backgrounds.

---

How to Use

1. Setup:

   • Install required libraries using:

     pip install ultralytics opencv-python numpy
     

   • Ensure the YOLO model file yolo11s-seg.pt is in the same directory as the scripts.
   • Place the video file (e.g., input3.mp4) in the directory.

2. Run the Scripts:

   • For single-frame segmentation:

     python yoloDetect.py
     

   • For 4-slice segmentation:

     python yoloSliceDetect.py
     

   • For 12-slice segmentation:

     python yoloSuperSliceDetect.py
     

3. Output:

   • Annotated video files will be saved as instance-segmentation3.avi.

---

Learn More

Visit Consulting Joe for more AI-powered solutions and tutorials.

---