SketchFormer: transformer-based approach for sketch recognition using vector images
- PDF / 1,402,934 Bytes
- 17 Pages / 439.642 x 666.49 pts Page_size
- 33 Downloads / 195 Views
SketchFormer: transformer-based approach for sketch recognition using vector images Anil Singh Parihar1
· Gaurav Jain1 · Shivang Chopra1 · Suransh Chopra1
Received: 26 February 2020 / Revised: 10 August 2020 / Accepted: 9 September 2020 / © Springer Science+Business Media, LLC, part of Springer Nature 2020
Abstract Sketches have been employed since the ancient era of cave paintings for simple illustrations to represent real-world entities and communication. The abstract nature and varied artistic styling make automatic recognition of these drawings more challenging than other areas of image classification. Moreover, the representation of sketches as a sequence of strokes instead of raster images introduces them at the correct abstract level. However, dealing with images as a sequence of small information makes it challenging. In this paper, we propose a Transformer-based network, dubbed as AttentiveNet, for sketch recognition. This architecture incorporates ordinal information to perform the classification task in real-time through vector images. We employ the proposed model to isolate the discriminating strokes of each doodle using the attention mechanism of Transformers and perform an in-depth qualitative analysis of the isolated strokes for classification of the sketch. Experimental evaluation validates that the proposed network performs favorably against state-of-the-art techniques. Keywords Sketch recognition · Transformers · Vector images · Deep learning
1 Introduction The ability to draw and comprehend eclectic notions through sketches reflects the intellectual capabilities of human beings. Drawing has long been a part of human behavior, Anil Singh Parihar
[email protected] Gaurav Jain [email protected] Shivang Chopra [email protected] Suransh Chopra [email protected] 1
Machine Learning Research Laboratory, Department of Computer Science and Engineering, Delhi Technological University, New Delhi, 110042, India
Multimedia Tools and Applications
mutating its utility from rock carvings in the ancient era, to blueprints of drafts in the modern age. Additionally, these sketch drawings assist humans in communication and creative design, such as art. Hence, intelligent machines must efficiently fathom the rapid proliferation of interaction with these human activities to be pervasive in all environments. Sketch analysis has been efficiently utilised in diverse domains like mathematics [18], chemistry [24], and electronics [23]. Further, sketch analysis spans a wide spectrum of applications, such as sketch segmentation [16, 19, 20, 35], image retrieval [12, 41, 43], and sketch recognition [11, 46, 48]. In this work, we focus on sketch recognition, which aims to identify human drawings and classify them into their respective categories. Although existing methods provide satisfactory accuracy performance for image recognition [17], only a few approaches have addressed the issue of identifying human drawn illustrations [27, 28, 45]. Sketch recognition was first introduced by Sutherlan
Data Loading...
