metadata
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
from transformers import pipeline
classifier = pipeline(
"audio-classification",
model="troyskie/Beewatch-hive-classifier"
)
result = classifier("sample.wav")
print(result)
Example Output
[
{'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.