Case Study

Education · Facial Recognition

Taking Attendance for a Whole Class in Three Minutes, With No Proxies

The Indian Institute of Information Technology, Kottayam (IIITK)

Project Snapshot

The Impact

3 min

To take attendance for an entire class, down from 10–15 minutes

95.6%

Recognition accuracy after refining the training data

2,500+

Student faces in the training dataset

7–12 min

Of teaching time returned to faculty in every class

01

Overview

The Indian Institute of Information Technology, Kottayam is an autonomous engineering institute in Kerala, working to international standards in research and teaching. Every class there began the same way — with a roll call — and three groups paid for it:

Manual attendance is not only slow; it is easy to cheat. When a friend can answer for an absent student, the record stops meaning anything — and it still takes ten minutes to produce.

Brain Station 23 developed a machine-learning facial recognition system tailored to IIITK, letting students record attendance through the LMS on their own phones, inside a set time window.

The result cut attendance to about three minutes per class, reached 95.6% recognition accuracy, and gave every lecturer back most of the time the roll call used to take.

The Indian Institute of Information Technology, Kottayam (IIITK)
02

Business Challenges

Attendance looks like a small problem until you multiply it by every class, every day, across a whole institute.

Four pressures shaped the system.

The system had to be faster than a roll call, harder to cheat than one, and invisible inside tools students already used.

Roll call eating into every lecture?

Bring it to a 30-minute call — we will come back with an approach and a scoped estimate.

03

What We Built

The design puts recognition on the student’s own phone and the record straight into the LMS, so attendance happens without the lecturer doing anything at all.

I

Automated attendance through the LMS

Students log in to the LMS on their phones, activate the camera, and their attendance is recorded automatically — no sheet, no roll call.

II

A model trained on 2,500+ faces

The recognition model was trained on a diverse dataset of more than 2,500 students’ faces, reaching 95.6% accuracy after the training data was refined, with real-time training improving it further.

III

Seamless LMS integration

Attendance lives inside the institute’s Moodle LMS, giving one platform for attendance and academic activity instead of two systems to reconcile.

IV

Time-windowed submission against proxies

Attendance can only be submitted during designated periods, which, combined with face matching, makes proxy attendance far harder.

V

A student experience people like

Students found the system convenient, and faculty no longer spend class time on administration.

The Indian Institute of Information Technology, Kottayam (IIITK)
04

The Results

The outcomes are measured in minutes per class, in records that can be trusted, and in what the institute now wants to do next.

Time and accuracy

Attendance now takes about three minutes for an entire class, down from 10–15 — giving faculty an extra 7–12 minutes of teaching in every session. Face matching eliminates proxy attendance and manual errors, making academic records reliable and engagement higher now that the lecture starts with teaching.

What comes next

IIITK plans to keep using the system and extend it to other areas of campus. Accurate identification also opens the door to safety uses, such as restricting access to dormitories and research labs to authorised people only.

The transferable lesson is that the time window matters as much as the model. Recognition proves who is present; a narrow submission window proves they were present at the right time.

Putting it inside the LMS students already used is what made adoption effortless. There was no new app to install and no new habit to learn.

The Indian Institute of Information Technology, Kottayam (IIITK)

Running a manual process that is slow and easy to game?

You might be taking attendance by hand, verifying identity with paper, or reconciling records between systems that do not talk. In every case, the first step is the same.

Book a 30-minute call with Brain Station 23. Bring the process and the systems it touches, and we will come back with a proposed approach, a delivery plan and a scoped estimate.

Roll call eating into every lecture?

Bring it to a 30-minute call — we will come back with an approach and a scoped estimate.

05

Related Case Studies