Assessfy Pvt. Ltd Beginner 5 milestones 100 marks

Handwritten Digit Recognition using CNN (MNIST + Indian Devanagari)

Target year: SE Sem 3-4 (Mini-Project-I) AICTE: 2 credits · ~50 hrs Bloom: Apply MU CBCS: AI403 Mini-Project 1A/1B

Real-world project · AICTE-aligned · AI-graded · Audit-ready certificate

Enroll as Student Mentor this project

Milestones

Available mentors

Enrolled students

Core skills

About this project

Train a CNN classifier on MNIST (10 English digits) + extend to Devanagari numerals (Hindi/Marathi). Compare a vanilla CNN vs a ResNet-style deeper net. Wrap in a Gradio web demo where users draw a digit and see the prediction live.

Course Learning Outcomes (CLOs):
CLO1: Apply CNN fundamentals to image classification.
CLO2: Compare model architectures empirically (vanilla CNN vs ResNet).
CLO3: Implement data augmentation pipeline.
CLO4: Deploy a live ML demo accessible via browser.

Industry/societal relevance: Devanagari OCR is unsolved at scale in India; this project is gateway-relevant for vernacular AI hiring at companies like Reverie, Karya, Sarvam AI.

Milestones

1. Dataset Setup

15 marks 5d

Load MNIST + Devanagari MNIST. Train/val/test split. Visualize samples + class balance.

2. Baseline CNN

20 marks 8d

2-conv + 2-FC CNN. Train on MNIST only. Report test accuracy >98%.

3. Cross-Script Extension

20 marks 10d

Re-train on combined MNIST + Devanagari (20 classes total). Show confusion matrix.

4. ResNet-style Comparison

20 marks 12d

Implement 4 residual blocks. Train + compare with baseline. Plot accuracy + parameter count.

5. Gradio Demo + Report

25 marks 15d

Live web demo: user draws a digit, sees top-3 predictions with probabilities. 6-page report.

Open internships using this project -->

Upcoming sessions

Session	Window	Enrolled
Handwritten Digit Recognition using CNN (MNIST + Indian D...	11 Jun 2026 to 10 Jun 2028	0

Skills you'll learn

PythonPyTorch/TensorFlowCNN architecturesImage preprocessingData augmentationGradio (live demo)Cross-script generalization

Tools used

Python 3.11PyTorch 2.x (or Keras)MNIST + Devanagari MNIST datasetsGradioGoogle Colab (free GPU)GitHub

Prerequisites

Python basics; intro to ML (lossoptimizertrain/test); basic linear algebra

Available mentors

Priyang Kumar

Free

You'll earn — Certificate (PDF)

AICTE-aligned Project Completion Certificate

A formal, audit-ready PDF certificate issued by Assessfy + your institute on successful completion. Includes AICTE credit hours, your evaluator's signature, and a QR code for third-party verification.

AICTE-aligned

Certificate of Project Completion

This is to certify that

has successfully completed the project

Handwritten Digit Recognition using CNN (MNIST + Indian Dev…

Scan to Verify

Date

Signature

Auto-issued on completion QR-verifiable

You'll earn — Digital Badge

Shareable LinkedIn / Resume Skill Badge

A compact, verifiable Open-Badges-2.0-compliant digital credential. Add to your LinkedIn profile, GitHub README, or resume in one click. Recruiters can validate authenticity via a unique URL.

Beginner

Handwritten Digit Recognition using C…

Assessfy

Auto-issued on completion One-click LinkedIn add