HSBD: Hu Sheep Behavior Dataset for Intensive Farming

280 images capture 6,766 instances of Hu sheep exhibiting four behaviors: standing, lying, eating, and drinking. Suoxiang Zhang created this dataset to address challenges in dense animal detection and class imbalance for computer vision models. The dataset was last updated on May 18, 2026.

Use Cases

• Training object detection models to identify sheep in dense, occluded farm environments based on the described image collection.
• Benchmarking model performance on imbalanced behavioral classification tasks using the four labeled categories.
• Developing lightweight vision models for edge deployment in farming scenarios, inspired by the KT-YOLO approach described.
• Studying multi-scale feature extraction methods for small object detection in agricultural settings.

Strengths

• Includes 6,766 annotated instances of Hu sheep across 280 images.
• Focuses on a specific, challenging agricultural scenario of dense sheep distribution and occlusion.
• Explicitly addresses four distinct behavioral categories: standing, lying, eating, and drinking.

Limitations

• Row count and column-level documentation are absent; field semantics must be inferred after download.
• The dataset is very small at 5.5 KB, indicating limited scope.
• Data may reflect geographic or farm-type bias inherent to its single-source collection.

Provenance

Source

Suoxiang Zhang via figshare.

Collection Method

Likely collected through image capture in an intensive Hu sheep farming environment.

Freshness

Last updated 2026-05-18 17:31:48; freshness should be verified.