MentorMindAI

Smart Video Evaluation and Accessibility Engine

MentorMindAI is an AI-powered backend system that evaluates teaching quality from recorded videos and converts educational content into accessible learning formats.
The platform uses ONNX-based machine learning models, FastAPI, and asynchronous processing to provide scalable, reproducible, and fair video evaluations.

---

Table of Contents

• Problem Statement
• Solution Overview
• Key Features
• Accessibility Modes
• System Architecture
• Project Structure
• Technology Stack
• Installation and Setup
• Running the Application
• API Endpoints
• Screenshots
• Contributors
• Future Scope
• License

---

Problem Statement

Teaching quality evaluation is often subjective and inconsistent. Additionally, standard video-based learning is not accessible to learners with visual, hearing, or cognitive challenges.

There is a need for:

• Objective teaching quality assessment
• Automated feedback for mentors
• Inclusive and accessible learning formats

---

Solution Overview

MentorMindAI provides:

• AI-based scoring of teaching videos
• Deterministic and reproducible evaluation metrics
• Automatic conversion of videos into accessible modes

---

Key Features

AI Video Scoring System

Each uploaded teaching video is evaluated using ONNX models across the following dimensions:

• Clarity
• Engagement
• Pace
• Filler Word Usage
• Technical Depth
• Weighted Overall Score

Each metric is processed independently to ensure fairness and transparency.

---

Accessibility Modes

Blind Mode

Generates audio narration describing both visual and spoken content.

Deaf Mode

Automatically generates subtitles using Whisper Speech-to-Text and exports .srt files.

Easy Mode

Produces simplified narration using text summarization and text-to-speech for better comprehension.

---

System Architecture