Automatically detect buildings using image and lidar data

Industry Energy and Utility

Specialization Or Business Function R&D, Exploration and Production, Engineering and Design, Physical Sciences

Technical Function Analytics (Machine Learning, Data Preparation)

Technology & Tools

COMPLETED Nov 05, 2014

Project Description

Company Information

We are a startup company out of MIT. Our customized multi-sensor hardware captures large amounts of data for millions of buildings. Through the use of robotics, computer vision, and big data analysis, the captured data is transformed into a meaningful information for our customers.

Data Challenge

1. Building localization

Data is recorded from thousands of buildings per evening. The vehicle-based data collection platform stores georeferenced long-wave infrared and near-infrared cameras images and 3D point clouds along with numerous other kinds of information. Determining which portions of the data localize individual buildings is a challenging computer vision and GIS effort. This challenge requires:

  • Using pre-existing training data to automatically determining the boundaries of buildings - specifically houses
  • Determine depth and dimensions of such buildings using calibrated LIDAR system
  • Isolating pixels of the structure and removing

Challenge may include mapping pixels between cameras using 3D features obtained from lidar data.

2. Building component identification

Long-wave infrared and near-infrared cameras record nearly one million images of energy loss scenes nightly. These co-collected images differ by scene coverage, but this is reconciled by time of collection and by overlapping features in 3D point clouds. Each building has a limited set of features (less than 20) that make up the most interesting parts of our energy analysis computations. Features include:

  • Windows
  • Doors
  • Walls
  • Soffits
  • Roofs
  • Etc.

Finding such features can be a challenge, depending on the type of construction, geography, image condition, and obstructions. All available data including images and point clouds may be needed to determine these features. Tens of thousands of features have been tagged for training data. Challenge may include mapping pixels between cameras using 3D features obtained from lidar data.


All of the deserialized information is stored in our cloud platform. Data is indexed by datetime and sensor. Both raw and normalized data is databased and queryable.

Project is hourly

This project seeks the highest value, not the lowest cost. We are gauging the responses based on experience and qualifications. The hourly rate range is $50-$150 per hour and will be based on merits.

Project Overview

  • Posted
    September 23, 2014
  • Planned Start
    October 20, 2014
  • Preferred Location
    From anywhere

Client Overview

Computer Vision
Image Analysis
Image Processing and Computer Vision
point cloud
Image Segmentation

Matching Providers