The same engine powers printed-sheet OCR, handwritten OCR, formula scoring, diagram comparison, and final report generation.
Professional grading experience for multimodal answer-sheet evaluation.
This platform turns your research pipeline into a product-style interface for OCR, formula interpretation, diagram analysis, scoring, and faculty-ready reporting.
Printed sheets with EasyOCR and handwritten sheets with Azure Document Intelligence.
Fast local SBERT flow plus Gemini-assisted grading when needed.
Text, formulas, and diagrams are processed as separate grading signals.
Core Product Modules
The interface is now focused around the exact capabilities needed to demonstrate the project as a polished academic product.
Document Capture
Uploads, rubric validation, PDF ingestion, and per-run workspace setup keep every batch traceable.
Recognition Layer
OCR, formula reading, and diagram region extraction convert answer sheets into structured evidence.
Scoring Engine
SBERT, CLIP, SymPy, and optional Gemini grading combine into rubric-aware marking.
Faculty Outputs
Live run tracking, ranking tables, downloadable PDF reports, and reusable JSON artifacts.
Simple Connected Workflow
Each stage feeds the next one so your guide can understand the system as a clean end-to-end product flow.
1. Upload Inputs
Select the ideal answer sheet, rubric JSON, and the student batch for the run.
2. Extract Signals
OCR reads the sheet while formula and diagram modules isolate non-text answer content.
3. Evaluate Answers
The selected engine compares each student response against the ideal answer under rubric rules.
4. Generate Reports
The pipeline stores JSON outputs, ranks students, and produces downloadable PDF reports.