🐾 Fursuit Head Detector (YOLO11m v2)

High-precision object detection model based on YOLO11m, specifically trained to detect fursuit heads in photos and videos. Built as part of an advanced computer vision pipeline for automated content processing.

📊 Dataset & Training Details

This model was trained on a meticulously curated dataset designed to eliminate false positives and handle complex real-world angles, lighting, and backgrounds:

Positive Samples: ~23,250+ manually labeled and verified images of fursuit heads.
Negative Samples: Over 75,000+ background images, general photography, animals, and non-fursuit content (not-furry and negative-dataset) to ensure robust rejection of non-target objects.
Image Size: 640x640 resolution.
Hardware Acceleration: Trained using PyTorch with CUDA on an NVIDIA GeForce RTX 5060 Ti (16 GB VRAM).

📈 Performance Metrics

The model achieves exceptional accuracy after extensive fine-tuning:

mAP50: $\approx 99.4%$
mAP50-95: $\approx 90.8%$

🚀 Quick Start (Python Inference)

You can easily load and use the model with the ultralytics library:

from huggingface_hub import hf_hub_download
from ultralytics import YOLO

model = YOLO(hf_hub_download("nexwayne/fursuit-yolo11m-v2", "best.pt"))

results = model("images.jpg")

results[0].save(filename="result.jpg")

Downloads last month: 114