IceKhoffi/chicken-object-detection-yolov11s

chbd-yolov11s-chicken-detector

This model is a YOLOv11s-based object detector specifically trained to identify individual chickens in farm environments. It is a core component of the "Chicken Health & Behavior Detection" multimodal project, aiming to provide visual insights for early disease detection and behavioral analysis in poultry farming.

Model Description

The chbd-yolov11s-chicken-detector is an object detection model trained using the Ultralytics YOLOv11s architecture. It is designed to accurately locate and classify chickens within images and video frames, serving as a foundational step for downstream tasks such as chicken counting, density estimation, tracking, and the analysis of anomalous visual behaviors related to health.

Training Data

This model was trained on the vision-object-detection datasets from chicken-health-behavior-multimodal.

The visual data within this dataset was curated from publicly available videos, specifically from the Kipster Farm YouTube channel. While Kipster Farm is located in the US, this data provides high-quality depictions of chicken behavior and general farm environments suitable for developing robust foundational detection models.

• Dataset Size : Total images used for training and validation: 24 (21 training, 3 validation)
• Annotations : Bounding boxes in YOLO format (class_id x_center y_center width height).
• Classes : Currently trained for a single class: 0.

Training Procedure

The model was trained using the Ultralytics YOLO framework (version 8.3.162)

• Model : yolov11s.pt
• Epochs : 100
• Image Size : 640x640 pixels
• Hardware : Training was performed on a Tesla T4 GPU with 15095MiB memory
• Optimizer : AdamW
• Python/Pytorch : Python 3.11.13, PyTorch 2.6.0+cu124

How to Use

You can load and use this model for inference with the Ultralytics YOLO library:

from huggingface_hub import hf_hub_download
from ultralytics import YOLO
import os

# Define each Hugging Face details
repo_id = "IceKhoffi/chicken-object-detection-yolov11s"
filename = "yolov11s.pt"
model_path = hf_hub_download(repo_id=repo_id, filename=filename)

# Load the trained model weights
model = YOLO(model_path)

# Perform inference on an image
results = model('path/to/your/image.jpg')

# Or on a video stream
# results = model('path/to/your/video.mp4', stream=True, save=True)

# Process results
for r in results:
    boxes = r.boxes # Bounding boxes
    masks = r.masks # Segmentation masks
    probs = r.probs # Probabilities

# Save inference results (if save=True in model call)
# Results are saved to runs/detect/predict by default.