YOLO Model Repository

A collection of YOLOv8 models for various object detection tasks, optimized for different deployment scenarios.

Available Models

Person Detection Models

• Nano: Lightweight model for person detection, optimized for edge devices

  • Format: OpenVINO
  • Size: 640x640
  • Task: Detection

• Small: Better accuracy for person detection with reasonable performance

  • Format: OpenVINO
  • Size: 640x640
  • Task: Detection

Vehicle Detection Models

• Small: Model for accurate vehicle detection
  • Format: OpenVINO
  • Size: 640x640
  • Task: Detection

Indonesian ID Card (KTP) Detection

• Nano: Specialized model for detecting Indonesian ID cards
  • Format: PyTorch (.pt)
  • Size: 640x640
  • Task: Detection

Model Architecture

These models are based on YOLOv8, the latest version of the YOLO (You Only Look Once) family of real-time object detectors. Key features:

• Anchor-free detection system
• Improved backbone and neck architecture
• Better performance-to-size ratio than previous YOLO versions

Model Performance (Still dummy, pls update the real ones later)

Model	Size	[email protected]	Inference Time (CPU)	Inference Time (GPU)
Person Nano	640x640	0.85	15ms	5ms
Person Small	640x640	0.89	25ms	8ms
Vehicle Small	640x640	0.87	28ms	9ms
KTP Nano	640x640	0.92	18ms	6ms