Skip to content

Streamlit Apps

ASCOTA ships with a set of Streamlit-based demo applications.
These apps provide a simple UI for archaeologists and researchers to test the core functions interactively — no coding required.

Each app wraps functions from ascota_core to demonstrate the Phase 1 pipeline (capabilities for segmentation, scale estimation, and color correction).

How to run

From the project root:

# Activate environment first
source .venv/bin/activate   # Windows: .venv\Scripts\activate

# Then run any of the apps:
streamlit run tests/streamlits/color_correct_streamlit.py
streamlit run tests/streamlits/scale_streamlit.py
streamlit run tests/streamlits/segment_streamlit.py
streamlit run tests/streamlits/color_clustering_streamlit.py
streamlit run tests/streamlits/color_clustering_lab_streamlit.py
streamlit run tests/streamlits/decoration_classification_streamlit.py
streamlit run tests/streamlits/type_classification_streamlit.py
streamlit run tests/streamlits/type_pottery_non_pottery_streamlit.py
streamlit run tests/streamlits/texture_clustering_streamlit.py

Open your browser at http://localhost:8501.

Available Apps

🎨 Color Correction

File: color_correct_streamlit.py

  • Detects color reference cards in input images.
  • Applies color correction to normalize pottery sherd images to a selected target image.
  • Useful for ensuring consistent analysis across lighting conditions and excavation seasons.
  • Allows testing of various color correction/transformation algorithms.

🌈 Color Clustering

File: color_clustering_streamlit.py * Clusters images based on dominant color features. * Groups photos taken under similar lighting conditions. * Helps archaeologists quickly identify and organize images by color similarity and lighting. * Useful for batch processing and allows us to select representative images for color correction.

📏 Scale Estimation

File: scale_streamlit.py

  • Detects measurement cards in excavation images.
  • Computes a pixels-per-centimeter ratio.
  • Estimates the surface area of sherds/pottery pieces from photos.
  • Provides archaeologists with approximate real-world measurements directly from images.

✂️ Segmentation

File: segment_swatch_streamlit.py

  • Runs segmentation models (RMBG-1.4 + OpenCV) to isolate sherds and measurement cards.
  • Produces clean masks and regions of interest (ROIs) for further analysis.
  • Segments & Classifies color cards, measurement cards in the image.
  • Generates an image swatch of the sherd.

🖌️ Color Clustering (Lab)

File: color_clustering_lab_streamlit.py

  • Runs Lab-summary-based clustering experiments for grouped pottery imagery.
  • Helps evaluate color-separation behavior before running full product workflows.

🧩 Decoration Classification

File: decoration_classification_streamlit.py

  • Demonstrates decoration classification over segmented sherd images.
  • Useful for testing model outputs and confidence behavior quickly.

🏺 Type Classification

File: type_classification_streamlit.py

  • Demonstrates multi-stage pottery type classification.
  • Useful for validating labels and confidence distributions on sample sets.

🏷️ Pottery Gate

File: type_pottery_non_pottery_streamlit.py

  • Runs pottery vs non-pottery gating for quick dataset screening.
  • Useful as a pre-step before detailed type classification.

🪨 Texture Clustering

File: texture_clustering_streamlit.py

  • Groups images by texture similarity using extracted surface features.
  • Helps inspect texture-driven grouping before downstream labeling.

Next Steps

These apps serve as experimental prototypes for Phase 1 of the ASCOTA pipeline. In later phases, additional apps will demonstrate:

  • Classification (Phase 2): automatic labeling by type, color, texture, and decoration/pattern.
  • Full Workflow (Phase 3): an end-to-end UI for archaeologists to process excavation photos.