A Pattern Recognition Method for Scene Sketch Images Using Deep Learning and Its Evaluation

Many studies on handwritten figure recognition have been published so far. For handwritten graphics in the fields of science and engineering, clear drawing rules are often clear, that is the shapes of the elements are predetermined. On the other hand, there is no drawing rule for sketches such as scene sketches; this makes it difficult to design a feature extractor. Based on this background, in recent years, research on recognition models using deep learning for sketch images has been reported. However, the recognition rate of all the previous research results are about 70% or less. Therefore, in this paper, we propose a new CNN model for recognizing landscape sketch images, and verify its effectiveness by computer experiments. Since there is no benchmark data for landscape sketch images, it cannot be compared with the results of previous studies, but the recognition rate of the CNN model proposed here is about 80%, which is higher than that of previous studies.