Automatically building appearance models from image sequences using salient featuresLocating salient object features