InterDrone 2019

Sessions Speakers Exhibitors Products Press Releases Floor Plan Buzz

Lies, Damned Lies, and Accuracy Metrics (Room Amazon Q-T)

06 Sep 19
9:00 AM - 10:00 AM

Tracks: Agriculture, Business, Cinema and Photo, Construction, Development and Hardware, Emergency and SAR, Energy, Fire, Infrastructure, Inspection, Law Enforcement, Public Safety, Surveying and Mapping

Speaker(s): Joe Morrison
This session is a beginner’s guide to:
  • Gaining fluency in common statistics used to describe machine learning models and their limitations
  • Understanding how they can be manipulated, both intentionally and unintentionally
  • Seeing some real examples of accuracy metrics done right and wrong
Advances in machine learning technology, especially the sub-field of deep learning, have dramatically expanded the number of use cases that drones can be applied to effectively. Whether the challenge is counting trees in an orchard, calculating the volume of a stockpile, or estimating the rust on a water tank, some vendor somewhere is likely using machine learning to expedite or automate that task. However, describing the accuracy of a machine learning model is far less straightforward than many of these vendors make it seem. In practice, statistical measures used to evaluate these models are often more misleading than they are helpful. This session will use real examples of drone projects that leveraged machine learning to illustrate how different evaluation techniques can be used to tell different stories about the same data, and to suggest best practices for portraying the accuracy of a model…accurately.