INNER CODE UNIT · Python

batch_incomplete

zuruoke/watermark-removal · inpaint_model.py:151

        batch_incomplete = batch_pos*(1.-mask)
        if FLAGS.guided:
            edge = edge * mask
            xin = tf.concat([batch_incomplete, edge], axis=3)
        else:
            xin = batch_incomplete
        x1, x2, offset_flow = self.build_inpaint_net(
            xin, mask, reuse=reuse, training=training,
            padding=FLAGS.padding)
        batch_predicted = x2
        losses = {}
        # apply mask and complete image
        batch_complete = batch_predicted*mask + batch_incomplete*(1.-mask)
        # local patches
        losses['ae_loss'] = FLAGS.l1_loss_alpha * tf.reduce_mean(tf.abs(batch_pos - x1))
        losses['ae_loss'] += FLAGS.l1_loss_alpha * tf.reduce_mean(tf.abs(batch_pos - x2))
        if summary:
            scalar_summary('losses/ae_loss', losses['ae_loss'])

View source record →