Objective To appraise facial skin's quality conveniently and correctly.Methods The quality appraisal model for facial skin image was established based on improved BP nerve network algorithm,and the model was trained with the intrinsic pattern of the training sample,then the appraisal result was output after inputting indexes on the skin.Results For 23 sets of samples,the appraisal accuracy reached 95%.Conclusion The network model,with a high accuracy and a low error rate,is applicable to facial skin image quality appraisal.