ShadeWise
ShadeWise is a teacher-centred OMR workflow for schools that use ordinary photographed answer sheets rather than dedicated scanning hardware.
As a teacher in a secondary school in my country, marking students answers can be stressful, time-consuming and there can be mistakes, so I thought we have AI that can handle these despite not have the hardware in the school to mark OMR sheet, and then I tried with ChatGPT for one of the classes I teach and the result and it's accuracy were amazing. Then I was like how can I help me colleagues so they can benefit from this at little to not cost. The school already has a functioning portal of which some other features were built using codex, and then I started prompting with the GPT 5.6 model to build this feature from start to finish The model tuned the OpenCV, found the best configuration/caliberation fit and also allows room for review rather than assuming (this is the part I love most)
Build Week contribution
The existing school portal predates Build Week. ShadeWise is the new assessment-analysis extension built during the submission period with Codex and GPT-5.6. The Build Week work includes the Django assessment workflow, OpenCV calibration and extraction pipeline, photographed-sheet frame/serial-row handling, confidence-based human review, protected scan delivery, raw-score analysis, exports, and automated tests.
The current release deliberately keeps raw MCQ results separate from the report-card exam score until a school sets its MCQ/theory/practical weighting policy.
Privacy and limitations
No external vision API, token quota, or dedicated scanning hardware is needed for extraction. The current version is tuned for a fixed bracket-box answer-sheet template; later work will support additional templates, bulk uploads, QR/student identification, annotations, and theory marking.
Log in or sign up for Devpost to join the conversation.