It’s Not Just about Accuracy: Metrics That Matter When Modeling Expert Sketching Ability
summary
A web‑based tutoring system called Sketchitivity collects stylus XY‑time data, resamples and segments strokes with ShortStraw, and uses 24 features (accuracy, smoothness, speed, planning) from a dataset of 2,000 sketches (20 novices, 4 experts) to train a classifier; complex forms like cubes reach 98% accuracy. Results show experts favor speed and smoothness over strict accuracy, break shapes into more strokes, and enable a holistic internal score that correlates with expert rankings.
AI-written summary · about
chapters (36)
- 0:00Introduction
- 0:35Benefits of drawing
- 1:10Design sketching
- 1:28Industrial design sketching
- 2:44Design Sketching Pedagogy
- 3:18SketchTivity
- 3:43Recognition
- 4:11Segmentation
- 5:01Accuracy
- 5:22Euclidean Distance
- 5:58Speed
- 6:15Smoothness
- 7:02Fluidity
- 7:22Planning
- 8:31Statistical Analysis
- 9:41Holistic Score
- 10:03Similarity Score
- 10:23Quality Score
- 10:49Accuracy Score
- 12:23User Interface
- 12:45Pressure
- 13:25Mechanics
- 13:46Music
- 14:31Sketching
- 14:42Our Mission
- 14:58Questions
- 15:25Beautification
- 16:11Can Machine Sketch
- 16:26Human Sketching
- 16:49Creative Systems
- 17:27Matching Styles
- 17:57Keep It Adaptive
- 18:25Do You Expect Users To Understand The Metrics
- 18:42Internal Metrics
- 18:53Drawing Metrics
- 20:03Dr Tracey Hammond
related talks
14mA Framework for Motivating Sketching Practice with Sketch-based Gameplay
39mPragmatic Sketching
49mPragmatic Sketching
48mKickstarting your Sketching and Visual Thinking skills
12mSquidgets: Sketch-based Widget Design for Scene Manipulation
07mScaffoldSketch: Accurate Industrial Design Drawing in VR
55mFrom Squiggles to Straight Lines: Sketch To Make Decisions & Get On With It
22mOne-Shot – A Quick Guide to Sketchnoting