The artistic style (or artistic movement) of a painting is a rich descriptor that captures bothvisual and historical information about the painting. Correctly identifying the artistic styleof a paintings is crucial for indexing large artistic databases. In this paper, we investigatethe use of deep residual neural to solve the problem of detecting the artistic style of apainting and outperform existing approaches by almost 10% on the Wikipaintings dataset(for 25 dierent style). To achieve this result, the network is rst pre-trained on ImageNet,and deeply retrained for artistic style. We empirically evaluate that to achieve the bestperformance, one need to retrain about 20 layers. This suggests that the two tasks are assimilar as expected, and explain the previous success of hand crafted features. We alsodemonstrate that the style detected on the Wikipaintings dataset are consistent with stylesdetected on an independent dataset and describe a number of experiments we conductedto validate this approach both qualitatively and quantitatively.