Skip to content

add SAM3/LangSAM install guide to README.md, extend model flag to run_pipeline.sh and new notebook for VLM-MASK-REASONER#18

Open
tharencandi wants to merge 3 commits intoNetflix:mainfrom
tharencandi:feat/stage3-segmentation-model-flag
Open

add SAM3/LangSAM install guide to README.md, extend model flag to run_pipeline.sh and new notebook for VLM-MASK-REASONER#18
tharencandi wants to merge 3 commits intoNetflix:mainfrom
tharencandi:feat/stage3-segmentation-model-flag

Conversation

@tharencandi
Copy link
Copy Markdown

Overview

This PR extends the stage3 model select to run_pipeline.sh in the VLM-MASK-REASONER pipeline, improves documentation for new users, and provides a new notebook specifically for the VLM-MASK-REASONER.

Changes Made

1. Stage 3 Segmentation Model Selection

  • Added --stage3-segmentation-model flag to run_pipeline.sh supporting both sam3 and langsam

2. Enhanced Documentation

  • Updated VLM-MASK-REASONER/README.md with:
    • Installation instructions for both SAM3 and LangSAM
    • Python version requirements and authentication details

3. Google Colab Notebook

  • Added notebook-vlm-mask-reasoner.ipynb demonstrating the full pipeline:
    • Setup and installation cells
    • SAM2 checkpoint download instructions
    • Configuration creation (manual and GUI-assisted options)
    • Pipeline execution (single command or individual stages)
    • Visualization and results download
    • Uses the lime sample from the repository

4. Code Cleanup

  • Removed unused code from stage3a_generate_grey_masks_v2.py

Testing

  • Tested full pipeline execution with LangSAM on Colab

- Add --stage3-segmentation-model flag to run_pipeline.sh (sam3|langsam)
- Document SAM3 and LangSAM installation requirements
- Clarify Stage 1 uses SAM2, Stage 3a uses SAM3/LangSAM
- Update VLM-MASK-REASONER README
@tharencandi tharencandi changed the title Add SAM3/LangSAM support and notebook for VLM-MASK-REASONER add SAM3/LangSAM install guide to README.md, extend model flag to run_pipeline.sh and new notebook for VLM-MASK-REASONER Apr 15, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants