Adaptive Sketchy Shape Recognition Based on SVM Incremental Learning
Zhengxing Sun, Lisha Zhang · 2005
This paper presents a strategy of adaptive online sketchy shape recognition. The inputting sketchy shapes are recognized online by means of a modified Support Vector Machine (SVM) incremental learning classifier. All classified results evaluated by user are collected and some important samples are selected according to their distances to the hyper-plane of the SVM-classifier. The classifier can then do incremental learning quickly on the newly added samples, and the retrained classifier can be adaptive to the user’s drawing styles. Experiments show the effectiveness of the proposed method.