IRIS

I built IRIS to solve a simple problem: students taking attendance for their friends (proxy attendance) and uploading forged certificates for attendance exceptions. IRIS stands for Intelligent Recognition System. It is an attendance platform that uses face matching, dynamic codes, and document checks to verify attendance.

Preventing Cheating

• ›Selfie Checks: When verifying, students take a live selfie. The system uses a face-matching model (VGG-Face) to compare it with their registered photo. It also uses texture checks to make sure they are taking a live photo, not just holding up a picture of a friend on another phone.

• ›Rotating QR Codes: Instead of a static code that students can copy and text to their friends at home, the teacher displays a QR code that rotates every 10 seconds. The student must scan it live in the classroom.

• ›Radar Scanner: Teachers can run a "Radar Loop" that automatically scans the room using a webcam to log multiple students quickly. To prevent students who show up late from logging attendance, the loop shuts down after 30 minutes.

Checking Document Fraud

If a student misses class and uploads a medical certificate, the system runs two checks. First, it uses Error Level Analysis (ELA) to see if the certificate was edited or photoshopped (by looking for differences in image compression levels). Second, it uses text recognition (OCR) to check if the name and dates on the certificate match the student's records.

Q&A Chatbot

We also integrated a simple chatbot that lets students ask questions about their attendance logs or college rules. The chatbot uses a local database (ChromaDB) to pull context from student handbooks, responding to queries in under a second.