Inspiration
Every day, engineers and technicians spend hours manually inspecting cars — from Formula 1 race cars to sedans rolling off factory lines — looking for scratches, dents, or subtle wear that could affect performance, safety, or brand trust. Even the most skilled human eyes can miss tiny defects.
What it does
-Detect tiny changes in vehicles or automotive components over time. -Classify the type of change: cracks, dents, paint fading, logo misplacement, or component wear. -Explain why the change matters using visual overlays and heatmaps. -Quantify changes with a degradation Index and track trends with temporal Drift Maps. -Support privacy preserving collaboration via Federated Learning.
How we built it
Image Processing using OpenCV + CLAHE for lighting correction and contrast enhancement. AI Model: CNN or Vision Transformer to detect and classify visual changes. Explainability: Grad-CAM/LIME to highlight regions of concern. Backend: Flask or FastAPI to handle image uploads and processing. Frontend: React.js or Streamlit for a dashboard showing before/after images, heatmaps, and temporal drift graphs. Database: MongoDB for storing images, results, and degradation history. Optional Add-On: Federated Learning using Flower/PySyft for collaborative, privacy-preserving training.
Log in or sign up for Devpost to join the conversation.