… We find that for both AI Sketch-a-Net and human judging, Pixelor … recurrent neural network
(RNN) that takes in a vectorized … sketch competitively with humans and often beat humans in …

… AI-Sketcher, a hybrid deep learning model that automatically … autoencoder based on a
convolutional neural network (CNN) to … Sketch-a-net that beats humans. arXiv preprint arXiv:…

… presents a sequential dual recurrent neural networks (SD-… proposed method with Sketch-A-Net
architecture coupled both … of 250 categories compared with human and Sketch-A-Net [39] …

… a new deep neural network, named Sketch-a-Net, to extract deep features. Sketch-a-Net …
Moreover, the performance of deep features is much higher than human when trained with …

… is now widely modeled by recurrent neural network (RNN). (v) … surpassing humans and
helped to popularize deep learning for … (d) Finally, an ensemble of Sketch-ANet were combined …

… We find that deep features as a perceptual similarity metric … further show that convolutional
neural networks trained on image … Sketch-a-Net: A deep neural network that beats humans. …

… [10] proposed an approach, namely, Sketch-a-Net (SN1.0), based on the utilization of the …
novel deep neural network model which is based on a Convolutional Neural Networks and the …

… recognition performance, even beats the human. It makes a … In sketch a net [19, 20]
approach, it shows very good … The reason is that CNN need large scale learning samples, …

… To exploit this piece of information, recurrent neural network … with LSTMs [31] and convolutional
neural networks as well [32]. … , “Sketch-a-net: A deep neural network that beats humans,” …

… and 20% of the amount of Sketch-A-Net and MobileNets. We also … , beat human on sketch
recognition task by constructing a … by designing a lightweight deep neural network which takes …