RankFrag: A Machine Learning-Based Technique for Finding Corners in Hand-Drawn Digital Curves

Abstract

We describe RankFrag: a technique which uses machine learning to detect corner points in hand-drawn digital curves. RankFrag classifies the stroke points by iteratively extracting them from a list of corner candidates. The points extracted in the last iterations are said to have a higher rank and are more likely to be corners. The technique has been tested on three different datasets described in the literature. We observed that, considering both accuracy and efficiency, RankFrag performs better than other state-of-art techniques.

Segmentation of Online Freehand Sketching Based on Speed Feature

“…Methods. Some stroke segmentation methods use machine learning techniques to determine the true corner points of a stroke from a set of candidate points [6,7,17]. ClassySeg [6,7] is a typical machine learning-based stroke segmentation technique to split hand-drawn pen strokes into lines and arcs.…”

“…However, ClassySeg relies on local decisions to construct a corner set, and it misses finding a globally optimal solution. RankFrag [17] uses machine learning technique to find corner points in hand-drawn strokes.…”

Shape Recognition and Corner Points Detection in 2D Drawings Using a Machine Learning Long Short-Term Memory (LSTM) Approach

“…There are two major approaches for Corner Detection: Traditional segmentation techniques, which typically rely on heuristic algorithms and empirically determined parameters, like curvature, and machine learning (ML)based stroke segmentation methods, which use various shape features to train a model. The latter is general and extensible compared to heuristic-based approaches [7].…”