SECURE AI GALLERY APPLICATION

An AI-powered gallery that automatically detects and protects sensitive photos like medical, banking, and ID images, ensuring other apps can access only safe public photos.

Comment

Inspiration

Today, when users grant gallery permission to an application, it gains access to all photos, including sensitive images such as medical reports, bank transaction screenshots, and identity documents. There is no intelligent system that automatically separates private images from normal photos. This creates a serious privacy risk. SecureLens AI was inspired by the need to protect user privacy at the gallery level without requiring users to manually manage sensitive photos.

What this project does

SecureLens AI is an AI-powered gallery prototype that automatically classifies images as PUBLIC or PRIVATE. The system analyzes images when they are uploaded, detects sensitive content such as medical, banking, or identity information, and protects those images while allowing safe photos to remain accessible. This ensures that other applications can access only public photos, reducing privacy exposure.

How we built it

The project was built as a web-based prototype. The frontend is developed using React and TypeScript, with a clean and responsive user interface. Images are uploaded and processed in the browser, then sent for AI analysis. We used Gemini AI Studio to design and test the AI reasoning logic. After extracting visible text and visual clues from images, Gemini analyzes the context and classifies each image. The core contribution of this project is prompt engineering, where we designed structured prompts to ensure fast, explainable, and reliable privacy decisions.

Challenges we ran into

One major challenge was handling multiple image uploads efficiently without slowing down the user interface or causing AI rate limits. We addressed this by using progressive background analysis and avoiding unnecessary deep analysis for clearly non-sensitive images. Another challenge was ensuring explainability, so users understand why an image is marked as public or private. This was solved by designing concise, user-friendly explanations from the AI.

What we learned

Through this project, we learned how multimodal AI models like Gemini can be used responsibly for real-world privacy problems. We gained hands-on experience with prompt engineering, scalable AI design, and building user-centric AI systems that prioritize transparency and trust rather than black-box decisions.

Future scope

SecureLens AI is a prototype demonstrating feasibility. In the future, this approach can be integrated at the operating system or app-permission level to provide stronger privacy protection. With on-device processing and deeper system integration, SecureLens AI could become a practical privacy layer for smartphones and digital platforms.

Built With

  • base64)
  • blob
  • googleaistudio
  • react-typescript-tailwind-css-google-gemini-3-flash-(via-google-ai-studio)-prompt-engineering-ocr-based-text-extraction-web-apis-(filereader
Share this project:

Updates