#### NOTE
This is a **community plugin**, an external project maintained by its respective author.
Community plugins are not part of FiftyOne core and may change independently.
Please review each plugin’s documentation and license before use.

<a href="https://github.com/danielgural/optimal_confidence_threshold" target="_blank">![GitHub Repo](https://img.shields.io/badge/GitHub-Repository-black?logo=github)</a>

# Optimal Confidence Threshold Finder

![Optimal Confidence](https://raw.githubusercontent.com/danielgural/optimal_confidence_threshold/main/assets/conf_output.png)

This plugin is a Python plugin that allows for you to find the optimal confidence threshold for deploying your models!

Find the magic number in one click!

# Installation

```shell
fiftyone plugins download https://github.com/danielgural/optimal_confidence_threshold
```

# Operators

## `optimal_conf_threshold`

Finding the optimal confidence threshold can always be tricky. One way to consider doing so is finding the number that maximizes your F1 score. This plugin implements a method of running an optimized version of FiftyOne’s `evaluate_detections` [method](https://docs.voxel51.com/api/fiftyone.core.dataset.html?highlight=evaluate%20detections#fiftyone.core.dataset.Dataset.evaluate_detections). This will calculate the F1 score at different thresholds until the best one is found! Here is how it works!

![Confidence Inputs](https://raw.githubusercontent.com/danielgural/optimal_confidence_threshold/main/assets/conf_inputs.png)

Input the following:

- Lower bound to search between
- Upper bound to search between
- Number of thresholds to check, this will increase complexity
- Ground Truth field
- Prediction field

The plugin will then sweep your range for the optimal confidence threshold until the answer is popped out!

![Optimal Confidence](https://raw.githubusercontent.com/danielgural/optimal_confidence_threshold/main/assets/conf_output.png)