Abstract:Sketch recognition remains a significant challenge due to the limited training data and the substantial intra-class variance of freehand sketches for the same object. Conventional methods for this task often rely on the availability of the temporal order of sketch strokes, additional cues acquired from different modalities and supervised augmentation of sketch datasets with real images, which also limit the applicability and feasibility of these methods in real scenarios. In this paper, we propose a novel sketch-specific data augmentation (SSDA) method that leverages the quantity and quality of the sketches automatically. From the aspect of quantity, we introduce a Bezier pivot based deformation (BPD) strategy to enrich the training data. Towards quality improvement, we present a mean stroke reconstruction (MSR) approach to generate a set of novel types of sketches with smaller intra-class variances. Both of these solutions are unrestricted from any multi-source data and temporal cues of sketches. Furthermore, we show that some recent deep convolutional neural network models that are trained on generic classes of real images can be better choices than most of the elaborate architectures that are designed explicitly for sketch recognition. As SSDA can be integrated with any convolutional neural networks, it has a distinct advantage over the existing methods. Our extensive experimental evaluations demonstrate that the proposed method achieves the state-of-the-art results (84.27%) on the TU-Berlin dataset, outperforming the human performance by a remarkable 11.17% increase. Finally, more experiments show the practical value of our approach for the task of sketch-based image retrieval.

Discovering Discriminative Patches for Free-Hand Sketch Analysis

Why Does the Sketch Look So Vivid?

Sketch-R2CNN: an Attentive Network for Vector Sketch Recognition

Free hand-drawn sketch segmentation

Coupling Deep Textural and Shape Features for Sketch Recognition.

Classified Segmentation of Online Complex Freehand Sketching

Segmentation of Online Freehand Sketching Based on Speed Feature

Sketch-Specific Data Augmentation for Freehand Sketch Recognition

Deep Learning for Free-Hand Sketch: A Survey

Self-Supervised Learning for Enhancing Spatial Awareness in Free-Hand Sketches

Segmentation of Online Sketching Using Geometric Feature

Judging and Grouping Methods for Online Overtraced Freehand Sketch

DifferSketching

A Preliminary Study On Extracting Objects In Sketches

Deep Self-Supervised Representation Learning for Free-Hand Sketch

Stroke-based semantic segmentation for scene-level free-hand sketches

Research on Freehand Sketchy Graphics Recognition

Iterative Collection Annotation for Sketch Recognition

Research on Recognition Effect of DSCN Network Structure in Hand-Drawn Sketch

Freehand Sketching Interfaces: Early Processing For Sketch Recognition

DeepFacePencil: Creating Face Images from Freehand Sketches.