# Confidence filtering

For every prediction made by the AI there will be a label that has the category name on it and below that label there will be a colored line that describes the confidence of the AI. Red means low confidence, yellow means medium confidence, green means high confidence.

![](/files/MGNauycGKg0wJ65tJEMa)

**Confidence Graph :** It shows the confidence of the AI's predictions in the 6 adjacent frames, this is useful to quickly find low confidence predictions that are confusing the AI and fix them. \
\
In the very early stages of annotation you may find the system predictions are noisy because the model is still confused. The best thing to do is to use confidence filtering to quickly clear the predictions and keep only the highest confidence ones. \
\
**Confidence threshold :** You can use the 'C' key to set the threshold level for confidence and all the predictions below that threshold will be removed once you start editing or confirm the frame. 

* All

![](/files/Pj85bxm3iHv8CgPlbW69)

* \>99%

![](/files/sQG0KtUubWA0yBvm8RA7)

* \>90%

![](/files/p1tkikREylPb1PJgZ6dV)

* \>80%

![](/files/mceryK2SXjmNvQw8qHlY)

* \>70%

![](/files/UzrzSTjjYqxLQjWwcg0H)

* \>60%

![](/files/niTx5Kso8wWrGynJuyIR)


---

# Agent Instructions: Querying This Documentation

If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.

Perform an HTTP GET request on the current page URL with the `ask` query parameter:

```
GET https://docs.lodestar.ai/dataset-management/video-annotator/confidence-filtering.md?ask=<question>
```

The question should be specific, self-contained, and written in natural language.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.