Monitor the Pace of Construction With AI

The client developed a product that monitors the progress of construction over time. To improve the accuracy of the product, they needed to annotate images of construction sites from various years.

In order to increase the accuracy of their product, the client needed their datasets to be annotated to train the machine learning algorithms. Such datasets would include images of how construction sites look today and how they looked a couple of years ago. Both of these datasets needed to be annotated so the system could compare them and measure the progress that was made, if any. Since there was a large amount of data that needed to be annotated, the client approached Mindy Support for assistance.

We assembled a team of 10 full-time data annotators to label the needed datasets. The client requested that we use QGIS to perform the annotation, which is a free and open-source tool specifically designed for annotating geospatial data. The output needed to be in GeoJSON format which is used for encoding a variety of geographic data structures.

We needed to label things like construction, parking, paths, roads, stadiums and paved and unpaved surfaces on the images. In total, we worked around 1,000 total hours to get the needed amount of data annotated within the specified time-frame.