source/modelling/dashboard/views.py

source-code dashboard visualization ui

File Path: src/modelling/dashboard/views.py

Purpose: high-level UI component rendering functions for different dashboard tabs.

Functions

render_metrics_view(y_true, y_pred, num_signs)

Displays classification metrics.

• Components:
  • Overall Accuracy (st.metric).
  • Classification Report (Precision, Recall, F1) as styled dataframe.
  • Confusion Matrix (Plotly heatmap).

render_error_view(y_true, y_pred, y_probs)

Analyzes misclassifications/

• Components:
  • “Top Confused Pairs” horizontal bar chart.

render_inspector_view(rnd_key, dataloader, model)

Interactive single-sample inspector.

• Features:
  • Random sample selection.
  • 3D Skeleton Animation (Plotly).
  • Real-time model prediction (if model loaded).
  • Confidence Bar visualization.

Calls:

• plot_3d_animation()
• shared_elements.get_visual_controls()

render_augmentation_view(rnd_key, dataloader)

Playground for testing data augmentations.

• Controls: Flip, Affine, Rotate, Scale, Shift.
• Visuals: Side-by-side comparison of “Original” vs “Augmented” 3D skeletons.

Calls:

• DataAugmentor

---

File Location: src/modelling/dashboard/views.py