Inspiration
RenoVision was inspired by the challenges homeowners face when dealing with property damage from natural disasters and daily wear and tear. These challenges disproportionately affect low- and middle-income families, adding to their financial stress.
Studies show that natural disasters can exacerbate poverty, making it harder for affected individuals to recover financially. Additionally, low-income homeowners are more likely to postpone essential maintenance due to budget constraints, which can lead to more severe damage and significantly higher repair costs over time. Moreover, assessing property damage typically requires professional evaluations, adding another expense and financial burden on homeowners. Without affordable and accessible assessment tools, people may overpay for services, delay necessary repairs, or underestimate the severity of damages, leading to decisions that escalate costs in the long run.
Our platform allows homeowners to quickly assess damage severity to make informed decisions without the reliance on costly professional interventions.
What it does
RenoVision is a web-based AI tool designed to help home and property owners assess damage to their properties, determine its cause and severity, and receive personalized repair cost estimates and suggestions. By analyzing uploaded images, the AI evaluates the type of damage, such as water leakage, fire, or mold, and provides insights into extent, from mild to severe. RenoVision empowers homeowners with direct, AI-driven insights so they can make faster, smarter repair decisions—all without needing an insurance adjuster or contractor for an initial assessment.
How we built it
We developed a custom AI model from scratch, gathering and labeling a diverse dataset of home damage images to train the model to recognize various types of disasters and issues, including water leakage, fire, and mold, while also assessing the severity of each damage type.
To make this technology accessible, we built a user-friendly website using Python, allowing users to upload images of their damaged homes.
Challenges we ran into
The greatest challenge throughout the process was finding a clean and useful dataset. While searching for image data from gettyimages, we encountered inconsistencies in image naming, and the database images were unlabeled. As a result, we had to manually gather data and label each image with its disaster type and severity before training the model.
Accomplishments that we're proud of
Unlike existing solutions designed for insurance adjusters and contractors, our website empowers homeowners with direct access to AI-driven damage assessment. By analyzing uploaded images, location data, and willingness to pay, our platform not only estimates repair costs but also considers real-world factors like material accessibility and labor difficulty.
What we learned
Throughout the development of RenoVision, we gained insights into both homeowner challenges and technical complexities in AI-driven damage assessment. Many homeowners, especially in lower-income groups, delay repairs due to financial constraints, leading to higher costs over time. Traditional expert evaluations add financial burdens, emphasizing the need for affordable, accessible assessment tools.
What's next for RenoVision
RenoVision is evolving to provide more precise and actionable insights for property owners. By introducing detailed damage classifications, such as pests and structural issues, the platform will enhance accuracy and usability. Leveraging past disaster data and market trends, it will offer context-aware repair strategies, making restoration efforts more informed and cost-effective. Additionally, a new predictive model using a LLM will estimate repair costs by factoring in location, material accessibility, and historical pricing trends, ensuring better accuracy in budgeting and planning. These advancements will empower users with personalized, data-driven renovation decisions.
Log in or sign up for Devpost to join the conversation.