Abstract:Sketches can be generated from color pictures or photos,but compared with pencil sketches,the outlines and shadows are not clear enough and the accuracy is not high.Due to the various types of sketch strokes,the shadow structure is complex.To this end,this paper proposes a method for converting face photos to pencil drawings,and designs and implements a new dual-branch training generative adversarial network model.The model has two branches,one contour branch improves the outline definition of the generated sketch image;the other shadow branch handles the texture and shadow parts of the sketch,keeping the sketch style.Finally,this paper conducts experiments on different types of sketch style conversion,and the results show that the proposed method has better sketch generation effect than existing methods.