1. Home
  2. Case Studies
  3. RoofQuote PRO: Ultimate AI-powered roofing solution

RoofQuote PRO: Ultimate AI-powered roofing solution

Data & Analytics Platform & Cloud Technology & SaaS AI & Gen AI Digital Experience Retail, Consumer & Hospitality AWS
Roofle ecommerce platform case study Roofle ecommerce platform case study

Client

Roofle Inc. is a leading American brand specializing in home roof repair, installation, and replacement. Founded by experts in marketing, sales, production, and software engineering, Roofle leverages advanced technology to transform the roofing industry, making roof purchasing faster and more convenient.

desktop-mobile-homeowner-instant-online-roof-quote-product-details-blue
1750432597742

Matthew McDaniel

Chief Marketing Officer and Product Owner

For years, I have had the privilege of working with Coherent Solutions on several applications, ranging from a cloud-based B2C e-commerce website to an industry disruptive SaaS platform that integrates bleeding edge tech such as machine learning. Their commitment to quality is unwavering and discernible in every project they undertake. The team not only excels in delivering world-class work but does so with remarkable consistency and professionalism. Their adept and methodical approach leaves no room for error, delivering outcomes that consistently exceed expectations. Truly, their unparalleled dedication to excellence sets them apart as a shining exemplar in the world of development.

arrow-forward-dark Connect With Our Experts

Business challenge

After two years of successful collaboration on the client's flagship product, RoofQuote PRO, the Coherent Solutions team identified an opportunity to further enhance the solution by integrating AI-powered features.

The idea was to advance the cost estimation process, pushing the boundaries of innovation and embracing the latest technology. This meant developing a service capable of generating detailed blueprints, facet counts, and precise area measurements of roofs using aerial imagery.

After careful consideration, Roofle enthusiastically embraced the idea, and the project moved forward.

roofle-power-of-instant-quotes

1.

From AI model training to finished product

2.

Scalable AWS infrastructure to accommodate user growth

3.

MLOps streamlined development pipelines to ensure reliability

Solution

The Coherent team carefully gathered all the necessary requirements, outlined a clear roadmap, and developed a service capable of extracting detailed blueprints, facet counts, and area measurements of roofs from aerial imagery. The solution involved several key steps:

  • Creating a custom semantic segmentation dataset: Ensuring the AI model had access to relevant, high-quality training data.

  • Training the AdaptNet model for accurate roof mask extraction: Employing the AdaptNet model architecture and training it extensively on the custom dataset to accurately extract roof masks from aerial imagery. 

  • Introducing an autoencoder to refine predicted footprints: Implementing an autoencoder neural network to refine predicted footprints generated by the segmentation model, reducing inaccuracies and improving the overall precision of the extracted roof shapes. 

  • Developing a module to convert model predictions to geographic coordinates: Designing and implementing a module to convert model predictions, such as segmented roof masks, into geographic coordinates, enabling seamless integration with mapping systems and geographical information, and precise location-based analysis. 

  • Obtaining vectorized roof representations by decomposing contours into edge vectors: Developing algorithms to break down roof contours, extracted from the segmented masks, into edge vectors, allowing for detailed, vectorized representations, suitable for further analysis and visualization of roof features. 

  • Developing a MobileNet classifier to assess mask quality, enhancing confidence scoring: Creating a MobileNet classifier to assess the quality of generated masks, evaluating factors such as completeness, accuracy, and alignment with aerial imagery. This enhanced confidence scoring ensured the reliability of the extracted roof features for downstream analysis. 

  • Creating an end-to-end solution for efficient roof analysis deployed on AWS: Integrating all components into an end-to-end solution for efficient roof analysis, leveraging AWS infrastructure for scalability, reliability, and cost-effectiveness. This deployment on AWS ensured seamless operation and accessibility of the AI-powered roof analysis service. 

Additionally, the team prioritized MLOps practices to maintain the solution’s efficiency and reliability. This included automating the continuous integration and deployment (CI/CD) pipeline using AWS CodePipeline and CodeBuild, as well as implementing ML service monitoring by extracting key metrics from daily logs and visualizing them on a user-friendly dashboard. 

Results

As the new AI features are deployed, the solution is already delivering tangible value. Roofle’s customers now gain a competitive advantage through an AI-powered tool that efficiently determines material waste factors. For contractor companies, this streamlines project planning and execution, while homeowners benefit from more accurate cost estimates and faster turnaround times.  

Overall, the integration of advanced AI models and algorithms has not only addressed Roofle's initial business challenge, it has also positioned the company at the forefront of innovation in the roofing industry.