humaisj/ring-detector-yolov8

Ring Detector YOLOv8 (Visual Ring Search)

This model is a fine-tuned YOLOv8 Nano object detection model designed specifically to detect rings in images. It serves as the "Stage 1" component of a Visual Search Engine for jewelry.

🚀 Project Overview

In a visual search pipeline, background noise (fingers, boxes, textures) can interfere with feature extraction. This model:

1. Identifies the ring in a raw image.
2. Crops the image to the bounding box of the ring.
3. Passes the clean crop to a background remover (rembg) and an EfficientNetB0 feature extractor.

📊 Training Details

• Base Model: yolov8n.pt (Ultralytics)
• Dataset: Custom dataset labeled via Roboflow (Dataset: 02_labeled_rings).
• Classes: 1 (Ring)
• Hardware: Trained on CPU/GPU via Jupyter Notebook.
• Training Code: Available in the project GitHub repository under notebooks/train_detector.py.

🛠️ How to Use

You can load this model directly using the ultralytics library:

from ultralytics import YOLO

# Load the model
model = YOLO('best.pt')

# Run inference
results = model.predict('your_image.jpg', conf=0.5)

# View results
for r in results:
    print(r.boxes)
    r.show()