EvaStudy

Gemin- based chat bot that remebers chat history
Correctly detecting happiness on human face.
Correctly detecting sadness on human face, suggesting break based on negative emotion displayed.
Home Page
Sign Up Page
Log In Page

Inspiration

We’ve all faced it: you sit down to study, and 10 minutes later you’re deep in a meme rabbit hole or doomscrolling TikTok. Or worse yet, you’ve actually managed to lock in and you find yourself getting angry and unproductive. In today’s world of constant digital distraction, staying focused isn’t just about willpower — it’s about having the right tools.

With online learning and self-study becoming the norm, we envisioned a smarter, emotionally-aware companion that supports—not scolds—you during your study sessions. That’s how EvaStudy was born.

📱 Detects distractions like phone usage in real time
💬 Recognizes emotional fatigue and offers personalized support
🎥 Boosts accountability with session recordings and insights

EvaStudy blends intelligent detection with compassionate feedback to help make every minute of studying more effective—and more human.

What it does

EvaStudy is an AI-powered web app that transforms how you study by combining multi-modal machine learning with conversational GenAI.

🎥 Monitor Sessions

Real-time phone detection using YOLOv5
Facial expression analysis to record emotion in a log and recommend breaks accordingly
Webcam recording for self-review and accountability

🤖 Deliver Smart Feedback via Chat Bot

Live nudges if you're distracted (via Gemini)
Encouragement when frustration is detected
Break suggestions based on emotional data

📹 Why record at 4× speed?
By saving sped-up versions of study sessions, EvaStudy makes it easier to review long sessions quickly, reflect on productivity patterns, and stay consistent. This also taps into the growing trend of “study-with-me” style content that helps users feel connected, motivated, and mindful—even during solo study time.

Business Model

We aim to empower focused learning—at scale.

Target Audience:

High school & college students
Working professionals seeking productivity tools
EdTech platforms looking to enhance engagement

Revenue Streams:

Individual Plans: Monthly/yearly subscriptions with core features
Institutional Licenses: Bulk discounts for schools and universities
Premium Add-Ons: Personalized coaching, advanced analytics
Partnerships: Integrations with learning platforms and referral models

Projections:

Year 1: 5,000 users | $300K revenue
Year 2: 15,000 users | $1M revenue | Breaking Even
Year 3: 40,000 users | $3M revenue

We're currently seeking $500,000 in funding to accelerate development, expand our reach, and build a community around focus and wellness in learning.

How we built it

Frontend:

Flask with Jinja2 templating
OpenCV video feed + custom CSS UI

Backend Highlights:

Object Detection: YOLOv5s model from Torch Hub identifies phones
Emotion Recognition: FER-2013 CNN via pre-trained model (AI-Gajendra)
Live Feedback: Gemini API crafts contextual nudges like “Put down your phone”
Cooldown Logic: Tracks recent emotions to prevent repetitive alerts
Video Recording: Captures sessions at 4× speed without audio (every 4th frame)
Storage: MySQL backend handles video metadata and user session logs

Stack Overview:

Python, TensorFlow/Keras, Torch, OpenCV
Google Gemini API, MySQL, Flask

Challenges

Balancing performance vs. accuracy in real-time video
Avoiding false positives in face and phone detection
Avoiding overrepresentation in emotion detection
Coordinating multiple AI models with shared session state
Designing interventions that support rather than annoy
Ensuring secure, isolated video recordings for each user and routing related to SQL, Flask and JavaScript