Digitizing Floor Plan Layouts using AI

Digitizing Floor Plan Layouts using AI

50 AI engineers collaborated in this high-impact 2-months innovation project to identify and construct digital objects from floor plans using computer vision.

 

The problem

To recognize floor plan elements in a layout requires manual labor to draw the different elements over the image. The goal of this project has been to improve the efficiency of this manual effort by automatically identifying the relevant types of objects present using state-of-the-art deep learning and computer vision approaches.

 

The project outcomes

The team built computer vision models to digitize the floor plan from architectural blueprints. The team successfully applied the following methods in achieving the tasks:

  • Object detection,
  • Image segmentation using Mask RCNN
  • Improved Optical character recognition (OCR) using the provided datasets,
  • Identifying languages other than English on floor plans

 

Data

Archilyse provided a large set of bitmap images of different sizes and dimensions, along with the bounding boxes of the relevant type of elements manually drawn. Examples of those are walls, columns, railings, kitchen furniture, shower, windows, doors, bathtub, bedroom area, kitchen area, etc.

 

Building an ML Model to Predict Future Infrastructure Needs of Africa for Policy Makers

Building an ML Model to Predict Future Infrastructure Needs of Africa for Policy Makers

The African Center for Economic Transformation (ACET) seeks to leverage AI to predict infrastructure needs within Africa to build a better future and create opportunities across countries.

 

The Problem

African governments are using significant portions of public budgets to finance infrastructure, but that infrastructure often responds to past or current needs, not future needs based on expected changes related to climate change, migration, urbanization, etc. Given limited fiscal space, African governments need to use all tools available to ensure the infrastructure being built today best serves the people of Africa for the next 50 to 100 years.

In this Omdena Challenge, a global community collaborated to predict the infrastructure needs of several African countries. We looked into various data sources such as satellite images, socio-economic data, climate, and topological data, population and demographic data, Google Trends, Google business data, social media data (to understand aspirations, needs, and sentiments of people living in the region), and other openly available data. The goal was to model the current situation, past temporal changes in population, infrastructure, etc., then predict future demands of infrastructure.

 

The Project Outcomes

As a team, the aim was to accomplish the following objectives:

  • Building one or multiple models for the future infrastructure needs of Africa (we will limit to selected groups of countries and certain types of infrastructure)
  • Modeling the aspirations of people in the given region of the world
  • Providing recommendations regarding verification approaches and networks to help scale to other countries

 

The Results

The Omdena team looked at the problem from different angles and used all tools to deliver the best solution. Like natural language processing, remote sensing, route planning, data analysis, and machine learning modeling. An interactive dashboard using Streamlit was implemented that makes it easier for the user to go around important and available data and predictions to make better policies and decisions.

A demo of the StreamLit dashboard

 

 

The dashboard gives visualizations and predictions to 5 main objectives:

  • Tweet analysis
  • African countries population
  • Electricity access
  • Distance calculations to vital amenities
  • Water stress index

 

To read more about the work done and methodologies used, check the articles attached below.

 

Building a Recommendation Engine for Sustainable Energy Solutions in Buildings

Building a Recommendation Engine for Sustainable Energy Solutions in Buildings

In this two-month Omdena Challenge, 50 technology changemakers have built a recommender engine to automate and simplify facility energy system upgrades. This will help to reduce energy wastage and carbon dioxide emissions while making buildings more efficient and sustainable. 

 

The problem

Existing buildings are the top consumer of energy, the top emitter of carbon dioxide emissions, and it takes 30-50 years for new energy efficiency solutions like LED lighting to fully penetrate our existing buildings One of the biggest challenges with upgrading our existing buildings with energy-efficient technology is identifying cost-effective solutions for individual buildings and asset portfolios. 

The current method is both time-consuming and expensive with sales reps to find customers, engineers manually inspecting the building’s existing energy systems, and then designing energy-efficient retrofit solutions, followed by a long sales cycle, low customer conversions, and no long term tracking of the retrofit project’s success. 

The main problem this project solved is the manual, time-consuming process to identify, prioritize, and recommend energy-efficient solutions for a particular building and customer. This will result in the building owner spending less on energy, spending less on their building upgrades, and emitting less carbon dioxide. Future iterations of the recommendation engine could also optimize for solutions that improve building occupant health and productivity which has social impacts or renewable energy, microgrid, water conservation, and other environmental solution adoption

 

The project outcomes

Retrolux has built a platform called Smart Energy scaleOS™ to automate and simplify facility energy system upgrades. You will build a project recommendation engine to the scaleOS™ platform that automatically analyzes individual buildings and asset portfolios to identify, prioritize, and recommend specific energy efficiency and other clean energy solutions that meet the facility owner’s project approval criteria.

The recommendation engine may analyze the building’s physical and energy characteristics, local and regional environment, utility tariffs and incentives, government regulations and incentives, current energy technology costs and performance specifications, and other available information.

The recommendation engine can then identify specific energy-efficient and clean energy solutions for individual buildings, provide means and methods to prioritize the solutions, and ultimately recommend cost-effective solutions for building owner approval based on the owner’s minimum return on investment, carbon reduction, or other goals. 

Our goal for this project is a robust overall recommendation engine architecture built specifically for at least one energy efficiency solution like LED lighting, lighting controls, rooftop unit optimization controls, smart motors, super-efficient warehouse freezer doors, etc.

 

Join our challenges here.

And find all our community benefits here.

Find more information on how an Omdena project works