| --- |
| 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. |
