An artificial neural network used to realize the approximating problem of the color appearance model (CAM) CIECAM02 in color management is demonstrated. GretagMacbeth ColorChecker Charts, which now axe widely used in calibration of digital camera, are chosen as samples to implement the forward and reverse color appearance models. When the predictive results are evaluated, for forward model, the output color appearance space is converted to the uniform color space based on CAM and is evaluated, while for reverse model, because the prediction precision is insufficient, we try to convert the color appearance space, which is the cylinder space, to the cube space similar to the red, green, and blue (RGB) space, and the results show that the precision is obviously improved. An artificial neural network used to realize the approximating problem of the color appearance model (CAM) CIECAM02 in color management is demonstrated. GretagMacbeth ColorChecker Charts, which now ax widely used in calibration of digital cameras, are chosen as samples to implement the forward and reverse color appearance models. When the predictive results are evaluated, for forward model, the output color appearance space is converted to the uniform color space based on CAM and is evaluated, while for reverse model, because the prediction precision is insufficient, we try to convert the color appearance space, which is the cylinder space, to the cube space similar to the red, green, and blue (RGB) space, and the results show that the precision is obviously improved.