SAS: Painting Detection and Recognition via Smart Art System With Mobile Devices

Abstract

Artwork recognition is an important research direction in the field of image processing. However, most of the current proposed methods are not designed for the demand of real-time analysis with mobile devices. Moreover, existing methods usually rely on high quality images and require large amounts of computing consumption. Based on the deep learning technology, in this paper, we propose a Smart Art System (SAS) with mobile devices. Our SAS mainly consists of two parts, i.e., painting detection unit and recognition unit. The detection module adopts a new painting detection algorithm called Single Shot Detection with Painting Landmark Location (SSD-PLL). SSD-PLL can effectively eliminate the influence of complex background factors on recognition. Considering the limited computing capacity of the mobile devices, our recognition module adopts a new ultra-light painting classifier. The classifier adopts MobileNet as the backbone and owns extra operation for Local Features Fusion (LFF). With our SAS, users can use mobile phone to take a photo of any paintings, then SAS would analyze the paintings and report the relevant information in real time. In order to validate the effectiveness of the proposed method, we have established two large scale image databases. The databases include 7,500 Traditional Chinese paintings (TCPs) and 8,800 Oil paintings (OPs), respectively. We evaluate our method and compare with the relevant algorithms, and our method achieves the highest performance and better real-time performance. Extensive experimental results on these databases show the effectiveness of the proposed algorithm.