Just How Accurate is Katapult Pro?
If you’re in the market for a new (or better) data collection method and you happen to come across our products or services, you might be wondering about the accuracy of our measurements. You may have other methods and products to compare us to, but real 1 for 1 comparisons between data collection methods are difficult to find. We have lots of numbers and complex math to show how accurate we are, but numbers don’t always get the picture across. Instead, we put our method through two days of testing. This allowed us to get a statistically significant number of pole photos and real-world measurements of those poles in order to test our method (and any other method). Our raw data from our testing is available below, but first we’ll explain how we tested our software and collection method.
We wanted to have as many experimental controls as possible, so we used a laser cutter to make our own calibration bracket (see image), accurate within 0.003”. Our custom bracket is 15” ± 0.003” from the top edge of the bracket to the bottom edge of the bracket, and the interior length from the two inside edges is 12” ± 0.003”. We attached the bracket to a streetlight pole in our parking lot and measured the distance from the bottom edge of the bracket to the ground (245.7” ± 0.125”). On to the experiment!
The Katapult Pro collection method involves two primary steps before delivering data–photo collection and data extraction. In order to best simulate a user new to the Katapult Pro method, we had 10 people take photos, each person having varied levels of experience in the field. Four of our photo collectors have done field/photo collection for years, three have a couple months of experience, two have only taken field collection photos once, and one person had no field collection experience at all. Each person took five full frame pole photos making sure the entire pole was in frame, and we uploaded these photos to Katapult Pro. We had one person extract the data from all 50 photos to ensure any differences in measurements were due solely to the photographers’ different angles to the pole and different distances from the pole when the photos were taken (Note: even though it looks like we only measure to the nearest 1”, we actually store the photo’s measurements down to the nearest 0.01”). What did the data show?
|Link to Full Chart||BEE||TIE||Pole Top||External Δ||Internal Δ|
|Real World (Actual) Height||245.7”||259.1”||291.5”||15”||12”|
|Average Measured Height||246.40”||259.85”||291.25”||14.94”||11.90”|
|Median Measured Height||246.45”||259.97”||291.35”||14.94”||11.92”|
|Mean Standard Dev.||0.4604”||0.5296”||0.8184”||0.1047”||0.0976”|
|Median Absolute Dev.||0.410”||0.405”||0.645”||0.080”||0.085”|
|Max Absolute Dev.||1.55”||1.65”||1.69”||0.27”||0.26”|
|Root Mean Square||0.8336”||0.9143”||0.8493”||0.1206”||0.1385”|
|Root Mean Square (%)||0.34%||0.35%||0.29%||0.80%||1.15%|
BEE = Height of Bracket’s Bottom Exterior Edge from Ground
TIE = Height of Bracket’s Top Interior Edge from Ground
External Δ = Difference between Bracket’s Exterior Edges
Internal Δ = Difference between Bracket’s Interior Edges
What do the numbers mean? When we deliver data to clients, there are two pieces of information in our deliverable that typically matter to them: absolute height measurements (heights measured from the ground) and relative height measurements (heights between things measured from the ground). For example, an absolute height would be used to determine the height of the pole top from the ground, and relative heights would be used to determine if a secondary power cable is greater than 40” from the highest communication attachment (for make ready clearances). We distilled all of the data down to find the Root Mean Square Percentage (the typical value used to determine data accuracy) of our absolute and relative heights. Our Average Absolute Height Accuracy is 99.50197%, and our Average Relative Height Accuracy is 99.02100%.
What does this mean for you? If you’re measuring a 50’ pole and are trying to find the pole top’s height above ground level, the value you measure is accurate to within 3”! This also means if you are doing make ready clearances for a utility or communications company, and you need to measure clearances of 40” or more, your relative heights (the differences between attachment heights) are accurate down to the nearest ¼”! How do other methods stack up? Go ahead, use this same experiment to test other data collection methods. We’re confident that Katapult Pro and the Katapult method is the most efficient data collection process out there. Want to put our method through the ringer yourself? Visit www.katapultpro.com today to find out more.