Skip to main content
Advertisement
Browse Subject Areas
?

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

< Back to Article

Fig 1.

A 1961 Sanborn map for Hanford Village in Columbus, Ohio.

(a) is the full Sanborn map sheet. Inset maps (b and c) show the details of the building-level information. Polygons shaded in yellow, pink, blue, and gray are the footprints of various buildings where the colors represent the materials used in building constructions. The abbreviated labels “D”, “F”, and “Auto” in each polygon represent dwelling, flat, and automobile, respectively. The numerals “1”, “2”, and “2 ½” in each building represent the number of stories.

More »

Fig 1 Expand

Fig 2.

A flowchart of the proposed workflow.

An orange-pink rectangle indicates input data, an open rounded rectangle indicates an operation, and a green rectangle represents intermediate or final output.

More »

Fig 2 Expand

Table 1.

Building construction materials and the colors that distinguish them on Sanborn maps.

More »

Table 1 Expand

Fig 3.

Pixel-based classification.

An orange-pink rectangle indicates input data, an open rounded rectangle indicates an operation, and a green rectangle represents intermediate or final output.

More »

Fig 3 Expand

Fig 4.

Postprocessing.

An orange-pink rectangle indicates input data, an open rounded rectangle indicates an operation, and a green rectangle represents intermediate or final output.

More »

Fig 4 Expand

Table 2.

Examples of letters and words as abbreviations of building utilizations.

More »

Table 2 Expand

Fig 5.

Network structure of Mask R-CNN.

More »

Fig 5 Expand

Fig 6.

Detecting building utilizations and numbers of stories.

An orange-pink rectangle indicates input data, an open rounded rectangle indicates an operation, and a green rectangle represents intermediate or final output.

More »

Fig 6 Expand

Table 3.

Number of pixels for each class in data sets P1 and P2.

More »

Table 3 Expand

Table 4.

Number of building objects for each class in data set O1.

More »

Table 4 Expand

Table 5.

Number of letters or words for each class in data sets U1 and U2.

More »

Table 5 Expand

Table 6.

Number of numerals for each class in data sets S1 and S2.

More »

Table 6 Expand

Fig 7.

Detected building footprints and their construction materials on a georeferenced Sanborn map.

(a) is a georeferenced Sanborn map, (b) shows the results of pixel-based classification, and (c) presents the vector building objects after post-processing. The colors used in the figure represent the different construction materials of the buildings, and their meanings can be found in Table 1.

More »

Fig 7 Expand

Table 7.

Results of the pixel-level accuracy evaluation.

More »

Table 7 Expand

Table 8.

Results of the object-level accuracy evaluation.

More »

Table 8 Expand

Fig 8.

Detected building utilizations (bounding boxes in red) and numbers of stories (bounding boxes in green) on Sanborn maps.

The text in white includes a detected abbreviation that denotes a building utilization or the number of stories, as well as a confidence score of the detection.

More »

Fig 8 Expand

Table 9.

Results of the accuracy evaluation for the detected building utilizations.

More »

Table 9 Expand

Table 10.

Results of the accuracy evaluation for the detected numbers of stories.

More »

Table 10 Expand

Fig 9.

Reconstructed historic neighborhoods of Hanford Village and Driving Park.

I-70 is colored orange, and the buildings that have been demolished are red.

More »

Fig 9 Expand