About the project
Checking descriptive answer sheets by hand takes hours after every unit test, and students usually get only a total with no explanation. This AI answer sheet evaluation Android project solves that. The teacher enters every question with a model answer and a marking scheme of key points. After the exam, the teacher photographs each student's handwritten pages in the app. AI reads the handwriting, splits it question-wise and suggests marks with the matched and missing points. Every mark stays in the teacher's hands: nothing reaches a student until the teacher reviews and publishes it.
How it works
- Admin adds classes and subjects, approves teachers and saves the Gemini API key
- Teacher creates an exam and adds questions with model answer, Bloom level and key points (AI can draft the key points)
- Teacher scans all pages of a student's sheet with the phone camera or gallery
- Gemini Vision reads the handwriting and splits it by question number (ML Kit OCR works offline)
- AI compares each answer with the marking scheme and gives marks in 0.5 steps, matched / missing points, similarity and a remark
- Teacher reviews each question, changes marks with a stepper and publishes the result
- Student sees question-wise feedback, downloads a PDF, requests re-check and practises weak topics with AI
Users
- Teacher: exams, questions and marking scheme, scanning, AI review, publishing, class analytics, re-check requests
- Student: results with feedback and PDF, re-check requests, AI practice with XP, streaks and badges, progress chart
- Admin (web): classes, subjects, teacher approval, students, exams, evaluations, re-evaluations, reports, notices, AI settings
Highlights
- 3 roles, 24 app screens, 14 admin panel pages, 35 PHP APIs, 14 MySQL tables
- Handwriting OCR with Gemini Vision plus offline ML Kit text recognition
- Rubric-based AI marking with an offline fallback (key point matching + cosine similarity)
- Class analytics with marks distribution chart and an AI-written class report (PDF)
- Online re-evaluation with an AI re-check for the teacher
- Validations in the app and in PHP (email, 10-digit mobile, password, marks steps, photo size)
- Hashed passwords, prepared statements, role and ownership checks on every API
- Sample data with demo logins for teacher, student and admin
What you get
- Complete Android Studio project (Java) and PHP source code
- MySQL database file with realistic sample data
- Project report in Word (49 pages) with DFD, ER and UML diagrams
- 2-page project abstract (Word + PDF)
- Project explanation with 20 viva questions
- Setup guide (Word + PDF) and README
Who can use this project
Students of BE / B.Tech Computer and IT, MCA, BCA, M.Tech and Diploma can submit it as a final year project. It is a good choice if you want an AI project in education, an OCR and handwriting recognition project, or an Android project with source code and documentation.
Frequently asked questions
Does it work without internet or a Gemini key? Yes. It switches to on-device ML Kit OCR and an offline marking checker automatically.
Can the teacher change the AI marks? Yes. AI marks are only suggestions. The teacher reviews every question and only the final marks are published.
Can it check diagrams and derivations? It evaluates the written text. For diagrams the teacher adjusts the marks on the review screen.
Is the Gemini API key free? A free key from Google AI Studio is enough for demos. It is pasted once in the admin panel.
