yolo26-onnx

YOLO26 ONNX Models

Complete collection of YOLO26 models in ONNX format.

📋 Overview

• Total Models: 19
• Source: Ultralytics/YOLO26
• Format: ONNX (opset 12)
• Input Size: 640×640
• Upload Time: 2026-02-24 16:50:57

🎯 Model List

Object Detection

• yolo26l.onnx (95.0 MB)
• yolo26m.onnx (78.2 MB)
• yolo26n.onnx (9.5 MB)
• yolo26s.onnx (36.5 MB)
• yolo26x.onnx (212.9 MB)

Instance Segmentation

• yolo26l-seg.onnx (107.1 MB)
• yolo26m-seg.onnx (90.2 MB)
• yolo26n-seg.onnx (10.7 MB)
• yolo26s-seg.onnx (40.0 MB)
• yolo26x-seg.onnx (240.0 MB)

Pose Estimation

• yolo26l-pose.onnx (99.4 MB)
• yolo26m-pose.onnx (82.6 MB)
• yolo26n-pose.onnx (11.6 MB)
• yolo26s-pose.onnx (39.9 MB)
• yolo26x-pose.onnx (220.0 MB)

Image Classification

• yolo26l-cls.onnx (53.9 MB)
• yolo26m-cls.onnx (44.4 MB)
• yolo26n-cls.onnx (10.8 MB)
• yolo26s-cls.onnx (25.7 MB)

🚀 Quick Start

import onnxruntime as ort

# Load model
session = ort.InferenceSession("zwh20081/yolo26-onnx/yolo26n.onnx")

# Prepare input (640×640 RGB, normalized 0-1)
# Inference...

⚙️ Technical Details

• ONNX Opset: 12
• Optimization: Simplified (simplify=True)
• Input: 3×640×640 (normalized 0-1)
• Device: CPU/GPU compatible

📄 License

• Models: AGPL-3.0
• Code: MIT

🔗 Links

• Ultralytics GitHub
• ONNX Runtime
• Original YOLO26

---

Repository created by: zwh20081