AeroVision

AeroVision automatically detects and visualizes design differences between F1 car images, helping engineers analyze updates in seconds.

Comment

Inspiration

In aerospace and motorsport engineering, design precision defines success. Teams frequently update aerodynamic components, yet manually comparing each iteration is time-consuming and error-prone. We were inspired by this challenge—how could AI help engineers instantly visualize design differences that might impact performance? That’s how the idea for the Visual Difference Engine (VDE) was born: an AI-powered tool that bridges design, analysis, and performance by automating the visual comparison process.

What it does

The Visual Difference Engine (VDE) identifies and highlights visual or structural differences between two images or 3D renders—such as between two versions of a car’s front wing or an aircraft’s fuselage model. It leverages computer vision to overlay and quantify changes, showing where and how geometry, materials, or surface features differ. This helps designers, engineers, and aerodynamicists instantly assess the impact of updates.

How we built it

We developed the system using:

OpenCV for image comparison and difference detection :Python for backend logic and AI model integration :Flask for a lightweight web interface :NumPy and Matplotlib for data processing and visualization

Our algorithm aligns the input images using ORB/SIFT feature matching and computes pixel-level and contour-based differences. To handle subtle modifications, we integrated a Siamese neural network, capable of detecting deep structural or design-level variations.

Challenges we ran into

A major challenge was ensuring consistent detection across images captured at different angles or under varying lighting conditions. To overcome this, we implemented normalization, adaptive thresholding, and perspective correction. Another key challenge was teaching the AI model to distinguish meaningful design changes from visual noise—achieved through careful dataset tuning and supervised training.

Accomplishments that we're proud of

We are proud of creating a working AI-driven visual comparison system that can detect even subtle structural differences between aerospace components with impressive accuracy. Our biggest achievement was successfully aligning and comparing images from different perspectives using computer vision—something that even professional software struggles with. Integrating a Siamese neural network for intelligent difference recognition was another milestone, allowing the tool to go beyond pixel comparison into design-level understanding. We also take pride in designing a simple yet powerful interface that makes advanced visual analytics accessible to engineers. Most importantly, we turned a complex engineering challenge into a real, usable solution that could genuinely accelerate design validation in the aerospace industry.

What we learned

We discovered how powerful computer vision can be when combined with domain-specific engineering insight. Our team learned to optimize image registration, develop AI-based similarity metrics, and manage the complexities of aerospace-scale image data. We also realized the importance of user interface design to make complex visual analytics intuitive.

What's next for AeroVision

Our next step is to extend the system for 3D CAD model comparison, allowing engineers to visualize internal or aerodynamic structural changes. We also plan to integrate a simulation link, where detected differences can be directly correlated to CFD or performance metrics. Ultimately, we envision VDE becoming an indispensable AI assistant for engineers working at the intersection of innovation, performance, and precision in aerospace design.

Built With

Share this project:

Updates

Submission history