README.md

---
pipeline_tag: token-classification
tags:
- drone-forensics
- event-recognition
license: mit
language:
- en
base_model:
- FacebookAI/roberta-base
library_name: transformers
---

# ADFLER-roberta-base

This is a [roberta-base](https://huggingface.co/FacebookAI/roberta-base) model fine-tuned on a collection of drone flight log messages: It performs log event recognition by assigning NER tag to each token within the input message using the BIOES tagging scheme.

For more detailed information about the model, please refer to the Roberta's model card.

<!--- Describe your model here -->

## Intended Use
![Description of Image](./concept.png)
- Use to split log records into sentences as well as detecting if the sentence is an event message or not.
- This model is trained diverse drone log messages from various models acquired from [Air Data](https://app.airdata.com/wiki/Notifications/)
## Usage (Transformers)

Using this model becomes easy when you have [transformers](https://www.SBERT.net) installed:

```
pip install -U transformers
```

Then you can use the model like this:

```python
>>> from transformers import pipeline
>>> model = pipeline('ner', model='swardiantara/ADFLER-roberta-base')

>>> model("Unknown Error, Cannot Takeoff. Contact DJI support.")

[{'entity': 'B-Event',
  'score': np.float32(0.9991462),
  'index': 1,
  'word': 'Unknown',
  'start': 0,
  'end': 7},
 {'entity': 'E-Event',
  'score': np.float32(0.9971226),
  'index': 2,
  'word': 'ĠError',
  'start': 8,
  'end': 13},
 {'entity': 'B-Event',
  'score': np.float32(0.9658275),
  'index': 4,
  'word': 'ĠCannot',
  'start': 15,
  'end': 21},
 {'entity': 'E-Event',
  'score': np.float32(0.9913662),
  'index': 5,
  'word': 'ĠTake',
  'start': 22,
  'end': 26},
 {'entity': 'E-Event',
  'score': np.float32(0.9961124),
  'index': 6,
  'word': 'off',
  'start': 26,
  'end': 29},
 {'entity': 'B-NonEvent',
  'score': np.float32(0.9994654),
  'index': 8,
  'word': 'ĠContact',
  'start': 31,
  'end': 38},
 {'entity': 'I-NonEvent',
  'score': np.float32(0.9946643),
  'index': 9,
  'word': 'ĠDJ',
  'start': 39,
  'end': 41},
 {'entity': 'I-NonEvent',
  'score': np.float32(0.8926663),
  'index': 10,
  'word': 'I',
  'start': 41,
  'end': 42},
 {'entity': 'E-NonEvent',
  'score': np.float32(0.9982748),
  'index': 11,
  'word': 'Ġsupport',
  'start': 43,
  'end': 50}]
```

## Citing & Authors
```bibtex
@misc{albert_ner_model,
  author={Silalahi, Swardiantara and Ahmad, Tohari and Studiawan, Hudan},
  title = {RoBERTa Model for Drone Flight Log Event Recognition},
  year = {2024},
  publisher = {Hugging Face},
  journal = {Hugging Face Hub}
}
```
<!--- Describe where people can find more information -->