White balance is an essential function for electronics with camera.Ho wever,the input images are indistinguishably processed by traditional methods,which decrease the processin g efficiency and image quality seriously.Therefore,the intelligent classifier is introduced to traditional wh ite balance algorithm using deep convolution neural network,judging whether the input image is polluted and proc essing those with color cast. There are three contributions in this paper:(1) The thought of processing after judging is firstly proposed. Specifically,only the image with color cast is processed while the normal image is output directly; (2) A valid training and validating dataset for color-cast image classification is construc ted by the proposed three random number method,which contains 34250images in the training dat aset,and the generation of color-cast image is highly random; (3) The open source VGG16model is optimized by learning rate dec ay and cross entropy loss function,and a fast-convergence and high-precision two-class classifier is t rained.The experiments show that the recognition precision for the normal image and the color-rendition image is 0.89and 0.92respectively.Intelligent white balance is realized by combining this classifier and histogram shift metho d.Artificial intelligence is introduced to traditional white balance in this paper,the processing efficiency and image quality is improved.