Study goals
Analyze the applicability of computational tools, with emphasis on Artificial Intelligence and computer vision algorithms, in the assisted generation of preliminary drafts of forensic engineering reports in cases of construction defects.
Relevance / originality
The proposal addresses the growing demand for speed, standardization, and reliability in forensic reports by leveraging emerging technologies to optimize the technical workflow, without replacing the expert’s critical analysis and subsequent validation.
Methodology / approach
The study adopts an exploratory and applied approach, involving the development of a Python-based prototype that integrates multimodal AI and the ORB algorithm, validated through testing with anonymized real-world forensic engineering report data.
Main results
The tests demonstrated the technical feasibility of the application, highlighting improvements in the speed of initial analysis, standardized report structuring, and the creation of a validated visual reference database, while maintaining expert review as a mandatory stage.
Theoretical / methodological contributions
The research proposes a hybrid model for assisted report generation, incorporating a human-in-the-loop framework and continuous feedback. It advances the interface between computing and legal engineering, offering methodological contributions to judicial forensic practice.
Social / management contributions
The proposal promotes increased efficiency in judicial proceedings by automating specific stages of forensic engineering workflows, contributing to the consolidation of a structured database of construction pathologies and strengthening the accuracy and standardization of future analyses.