Meeting Summation and Notes

Meeting Transcript Overview

Project Update

  • Confirmation that feedback was received regarding the project.

  • Inquiry if the feedback was emailed to the professor; clarification that Dave was responsible for sending it.

    • Individual responses were sent directly to Dave, who would forward them to the professor.

GitHub Access Discussion

  • Discussion led by Sudeep regarding GitHub access for the project team.

    • Suggestion for team members to use personal GitHub usernames instead of GMU email for access continuity.

    • Emphasis on giving usernames to facilitate smoother transitions and operations.

  • Offers to consolidate usernames into a single email to expedite the process.

AWS and Project Foundations

  • Mention of AWS in relation to GitHub, assuring the team that using AWS alongside GitHub would provide necessary resources.

  • Assurance from the speaker that the foundational setup for team projects would be established.

Provenance Tracking

  • Introduced a schema for tracking various data points related to project development:

    • Categories to track:

    • Producer of data

    • Type of data

    • Timestamp

    • Model schema

    • Parameters

  • Specific examples discussed included input and output for OCR pipelines and JSON formats including confidence scores and quality metrics.

  • Updates to the JSON file schema to automate the tracking mechanism.

    • Integration with OCR pipeline and Python normalization scripts predicated upon JSON.

  • Note that data would append to a CSV file for traceability.

Pipeline Diagram Overview

  • Presentation of a diagram that outlines the project pipeline:

    • Initial data input (text/images) -> extraction to JSON format -> JSON validation -> multi-file merging -> conversion to IFC -> final 3D modeling.

  • Example from the floor plan used to check information flow through the stages.

  • Use of Blender for 3D visualization and model rendering.

    • Challenge noted with rendering details accurately from JSON to IFC in Blender.

  • Initial results showcased, indicating a learning curve in getting accurate models from the system.

Reverse Engineering Discussion

  • Proposal to reverse engineer an existing IFC model to enhance schema creation:

    • Use information from IFC to inform JSON structure and improve accuracy for future models.

  • Discussion on obtaining a small IFC file for testing reverse engineering processes to establish effective schemas.

  • Clarification that the goal is a well-defined schema that can be used consistently across projects moving forward.

Feedback on Progress and Challenges

  • Positive reinforcement of progress made so far in modeling and file conversions.

  • Mention of the transformation from single walls to multi-wall structures during the modeling processes, indicating substantial growth or progress in the project.

  • Suggestions on what additional context could be included to improve JSON conversion accuracy, particularly around specifying elements like floors, ceilings, and details in the designs.

Multiple AI Models and Collaboration

  • Discussion on the new Google protocol for collaborative AI workflows: A to A (agent-to-agent) model allowing the use of multiple AI systems in tandem to strengthen results.

  • Emphasis on the potential benefits of using different AI models for specific tasks (such as extracting data vs final formatting) to create a superior output.

Administrative Notes

  • Mention of upcoming deliverables:

    • Interim Written Report: Summary of progress, challenges, technologies used, and future architecture plans.

    • Monthly Report: Similar to previous satisfaction reports, it should reflect scores and comments from the team.

    • Instructions on how to email reports to the professor for direct feedback acknowledgment.

  • Discussion regarding the final deliverables due in December and confirmation of last meeting date.

  • Questions raised surrounding the need for an IFC file and expectations around its release.

Next Steps and Closing Remarks

  • Team agreement on scheduling a final meeting on December 5, 2023.

  • Possibility of additional feedback before presentations scheduled around that time, with future work planned leading into the next academic semester (early January 2024).

  • Expressions of gratitude and good wishes for Thanksgiving and the holiday season.

Key Takeaways and Actions

  • Immediate Actions:

    • Email GitHub usernames to Sudeep.

    • Coordinate with team on feedback integration for interim reports summarizing progress and what’s needed for future success.

    • Schedule prospective time for future meetings related to presentations in early December.

    • Establish methodologies for validation processes in project outputs in preparation for the next semester deadline.