Add Prediction Confidence Score in ASL Recognition Results

[muskan-202]

Currently, the application displays the predicted sign/character, but users cannot determine how confident the model is about its prediction.

Adding a confidence score will improve transparency and help users understand when a prediction may be unreliable.

Proposed Solution
Display the model confidence percentage alongside the predicted character/word.
Add a confidence threshold warning for low-confidence predictions.
Show confidence in a user-friendly format (e.g., 92.4%).
Ensure the confidence score updates in real time with predictions.

#Benefits
Better user trust in model predictions.
Easier debugging and model evaluation.
Improved accessibility and learning experience.

#Acceptance Criteria
Confidence score is displayed with every prediction.
Score updates in real time.
Low-confidence predictions are highlighted.
UI remains responsive and readable.

[muskan-202]

Hi @Sant60 ,

I would like to work on this issue under GSSoC'26.

Approach:

1. Identify the prediction pipeline where model outputs are generated.
2. Extract prediction probabilities/confidence values.
3. Display confidence percentage in the UI.
4. Add visual indication for low-confidence predictions.
5. Test with multiple gestures to verify accuracy and responsiveness.

Kindly assign this issue to me.

Thank you!

[Nairayadav]

Hi @Sant60,

I would like to work on this issue under GSSoC'26.

I have reviewed the requirement and understand the need to display prediction confidence alongside the recognized ASL character. I can implement real-time confidence score updates, add a visual indicator for low-confidence predictions, and ensure the UI remains responsive and user-friendly.

Kindly assign this issue to me if it is available.

Thank you!