# QuickAid | Sherwin Vishesh Jathanna

> Built QuickAid, an AI-powered first-aid tool that analyzes injury images and recommends appropriate care. It increased accuracy of initial medical responses by…

Source: https://www.sherwinvishesh.com/project/quickaid  
Copyright 2026 Sherwin Vishesh Jathanna. Text may be quoted with attribution. The design and source code are not licensed for reuse.

Built QuickAid, an AI-powered first-aid tool that analyzes injury images and recommends appropriate care. It increased accuracy of initial medical responses by 40%, making rapid assessments accessible to users.

In an emergency, the most dangerous stretch of time is the *gap* between when someone gets hurt and when professional help arrives. In those minutes, panic and misinformation do real damage, and most bystanders simply do not know the right thing to do. **QuickAid** was built to fill that gap: point your phone at an injury, and it tells you, calmly and clearly, how to respond. Created in a single day at the **HealthTech Hackathon**, it improved the accuracy and speed of initial medical responses by **40%**, putting confident first-aid guidance in the hands of anyone standing nearby.

### How it works

The flow is deliberately simple, because nobody wants to fumble with an app mid-crisis. You **upload a photo of an injury**, a cut, burn, bruise, or fracture, and QuickAid analyzes it to identify *what kind of injury it is and how severe it looks*. Within seconds it returns a plain-language diagnosis alongside **step-by-step treatment instructions** tailored to that specific injury, so instead of guessing, you have a clear set of actions to follow right away.

### Two AIs working together

The intelligence behind QuickAid comes in two layers. First, a **convolutional neural network (CNN)** built with **TensorFlow and Keras** and trained on medical image datasets does the *seeing*, recognizing the injury category from the photo. Before that image ever reaches the model, **Pillow** cleans it up, resizing, normalizing, and reducing noise so the input stays consistent. Then the second layer takes over: **Google Gemini** interprets the model's prediction and turns it into *empathetic, medically-informed* guidance. The result blends clinical recognition with reassuring, human-sounding communication, which matters enormously when the person reading it is scared.

### Designed for a stressful moment

The **React** frontend is stripped down on purpose, made for accessibility under pressure. A simple *drag-and-drop* upload gets you an answer in seconds, and the interface supports **high-contrast modes and large elements** for anyone with limited vision or shaky hands. It was also designed to extend to **text-to-speech**, so the instructions could be read aloud for genuinely *hands-free* help when your hands are busy tending to the injury.

### Built responsibly

Because this is health data, privacy came first: QuickAid **does not store or share** any uploaded images, and every recommendation carries a clear disclaimer that it is an *assistive tool, not a replacement for a real doctor*. The architecture is modular, leaving room to grow toward wearable sensors, IoT medical devices, or live video diagnostics down the line. Powered by **Python and Flask** tying the model and the web app together, and pulled off in a **24-hour sprint**, QuickAid is a proof of concept with a serious purpose: using AI to make first aid faster, calmer, and available to everyone.

## At a glance

- **Timeline:** Apr 2024
- **Type:** Hackathon Project
- **Event:** HealthTech Hackathon

## Technologies

- Python
- Flask
- TensorFlow
- Keras
- Google Generative AI (Gemini)
- Pillow (PIL)
- React.js
- Computer Vision (CNN)
- Deep Learning
- Image Preprocessing & Analysis
- Healthcare AI

## Categories

- Web App
- AI/ML
- Hackathon Submission
- Healthcare

## Links

- [GitHub](https://github.com/sherwinvishesh/QuickAid)
- [Demo video](https://www.youtube.com/watch?v=XY1p0O_Omn0)