This dataset includes different videos of various broilers’ behaviors gathered to train a video classifier model. The collected videos were manually classified into three classes: preening, stretching, and normal. The preening class contains frames where a broiler preens its feathers with its beak; the stretching class includes frames where a broiler stretches one wing and one leg; and the normal class represents other broiler behaviors such as sitting, standing, feeding, and drinking. To enhance the accuracy of the video classifier model’s results, efforts were made to ensure that each video contained only one broiler. All videos were recorded from commercial farms using a top-view camera setup.

The dataset was created to support research in:

• Automated detection and classification broiler behaviors.

• Video recognition-based pipeline to detect and quantify the broilers’ behaviors.

• Welfare monitoring.

🗄️ Dataset Contents

This dataset contains annotated broiler videos classified into behavioral categories:

🗂️ Data Structure

📁 BehaviorRecognition/

│

└── 📁 Normal/

├── 📄 N0.mp4

├── 📄 N1.mp4

└── ...

└── 📁 Preening/

├── 📄 P0.mp4

├── 📄 P1.mp4

└── ...

└── 📁 Stretching/

├── 📄 S0.mp4

├── 📄 S1.mp4

└── ...

📑 Data Summary

📥 Dataset Download

You can directly download the dataset from here:

📚 Citation

If you use this dataset, please cite:

Nasiri, A., Zhao, Y. and Gan, H., 2024. Automated detection and counting of broiler behaviors using a video recognition system. Computers and Electronics in Agriculture, 221, p.108930. DOI: https://doi.org/10.1016/j.compag.2024.108930

🤝 Contributing

Contributions are welcome!