Anchor Free YOLOv8 outputs about ultralytics

[glenn-jocher]

Hello there! 👋

Great question! YOLOv8 moves away from the anchor-based predictions seen in earlier YOLO versions. Instead, it relies on an anchor-free approach. This means that rather than predicting multiple bounding boxes per grid cell based on predefined anchors, YOLOv8 directly predicts the bounding boxes without anchors. Each grid cell can predict a box based on the object's center falling within that cell, allowing for more flexible and potentially more accurate detections.

Here's a simple conceptual illustration:

# In earlier YOLO versions with anchors
# Each grid might predict something like this (simplified):
[
  {"x_center": 0.5, "y_center": 0.5, "width": 0.2, "height": 0.3, "confidence": 0.9},  # Anchor 1
  {"x_center": 0.5, "y_center": 0.5, "width": 0.25, "height": 0.35, "confidence": 0.85}, # Anchor 2
  {"x_center": 0.5, "y_center": 0.5, "width": 0.3, "height": 0.4, "confidence": 0.8}    # Anchor 3
]

# In YOLOv8 anchor-free approach
# A grid cell directly predicts:
{
  "x_center": 0.5, "y_center": 0.5, "width": 0.24, "height": 0.36, "confidence": 0.92
}

This shift enhances the model's ability to detect various sizes and shapes of objects more efficiently. Hope this clears up how YOLOv8 differs from previous versions in terms of bounding box predictions! If you have any more questions, feel free to ask. 😊

[github-actions]

👋 Hello @Mikael17125, thank you for your interest in Ultralytics YOLOv8 🚀! We recommend a visit to the Docs for new users where you can find many Python and CLI usage examples and where many of the most common questions may already be answered.

If this is a 🐛 Bug Report, please provide a minimum reproducible example to help us debug it.

If this is a custom training ❓ Question, please provide as much information as possible, including dataset image examples and training logs, and verify you are following our Tips for Best Training Results.

Join the vibrant Ultralytics Discord 🎧 community for real-time conversations and collaborations. This platform offers a perfect space to inquire, showcase your work, and connect with fellow Ultralytics users.

Install

Pip install the ultralytics package including all requirements in a Python>=3.8 environment with PyTorch>=1.8.

pip install ultralytics

Environments

YOLOv8 may be run in any of the following up-to-date verified environments (with all dependencies including CUDA/CUDNN, Python and PyTorch preinstalled):

• Notebooks with free GPU:
• Google Cloud Deep Learning VM. See GCP Quickstart Guide
• Amazon Deep Learning AMI. See AWS Quickstart Guide
• Docker Image. See Docker Quickstart Guide

Status

If this badge is green, all Ultralytics CI tests are currently passing. CI tests verify correct operation of all YOLOv8 Modes and Tasks on macOS, Windows, and Ubuntu every 24 hours and on every commit.