---
library_name: transformers
tags:
- audio-classification
- bee
- beehive
- ast
- transformers
- environmental-audio
---

# BeeWatch Hive Classifier

## Overview

BeeWatch Hive Classifier is an AI model that classifies beehive sounds into three categories:

- Healthy
- Warning
- Low Activity

The model is designed to assist beekeepers in monitoring hive health using audio recordings.

---

## Model Details

- Model: ASTForAudioClassification
- Base Model: MIT/ast-finetuned-audioset-10-10-0.4593
- Framework: Hugging Face Transformers
- Task: Audio Classification

---

## Labels

| Label | Description |
|--------|-------------|
| Healthy | Hive is functioning normally. |
| Warning | Hive exhibits unusual activity that may require inspection. |
| Low Activity | Hive activity is significantly reduced. |

---

## Dataset

The model was fine-tuned using a custom beehive audio dataset containing WAV recordings collected from different hive conditions.

Classes:

- Healthy
- Warning
- Low Activity

---

## Usage

```python
from transformers import pipeline

classifier = pipeline(
    "audio-classification",
    model="troyskie/Beewatch-hive-classifier"
)

result = classifier("sample.wav")
print(result)
```

---

## Example Output

```text
[
 {'label': 'Healthy', 'score': 0.96},
 {'label': 'Warning', 'score': 0.03},
 {'label': 'Low Activity', 'score': 0.01}
]
```

---

## Limitations

- Performance depends on audio quality.
- Excessive environmental noise may reduce accuracy.
- The model should assist, not replace, manual hive inspection.

---

## Intended Use

This model is intended for research and educational purposes in smart beekeeping and hive health monitoring.