Sketch-a-Net that Beats Humans

TL;DR

This paper introduces a specialized deep neural network for sketch recognition that outperforms human accuracy by incorporating sketch-specific features, multi-scale ensemble methods, and sequential encoding.

Contribution

The authors develop a novel multi-scale, multi-channel deep network architecture tailored for sketches, surpassing human performance and existing photo-based models in sketch recognition.

Findings

Achieves state-of-the-art accuracy on large sketch dataset

Outperforms humans in sketch recognition tasks

Efficient training with CPU-based implementation

Abstract

We propose a multi-scale multi-channel deep neural network framework that, for the first time, yields sketch recognition performance surpassing that of humans. Our superior performance is a result of explicitly embedding the unique characteristics of sketches in our model: (i) a network architecture designed for sketch rather than natural photo statistics, (ii) a multi-channel generalisation that encodes sequential ordering in the sketching process, and (iii) a multi-scale network ensemble with joint Bayesian fusion that accounts for the different levels of abstraction exhibited in free-hand sketches. We show that state-of-the-art deep networks specifically engineered for photos of natural objects fail to perform well on sketch recognition, regardless whether they are trained using photo or sketch. Our network on the other hand not only delivers the best performance on the largest human sketch dataset to date, but also is small in size making efficient training possible using just CPUs.