diff --git a/tools/eval.py b/tools/eval.py
index 16cfe532aae49ce98bc9503ca73e009bf206caa7..4afed469c875ef8d2200cdbfd89e5a8af4c6b7c3 100755
--- a/tools/eval.py
+++ b/tools/eval.py
@@ -47,6 +47,7 @@ def main():
         config['Architecture']["Head"]['out_channels'] = len(
             getattr(post_process_class, 'character'))
     model = build_model(config['Architecture'])
+    use_srn = config['Architecture']['algorithm'] == "SRN"
 
     best_model_dict = init_model(config, model, logger)
     if len(best_model_dict):
@@ -59,7 +60,7 @@ def main():
 
     # start eval
     metirc = program.eval(model, valid_dataloader, post_process_class,
-                          eval_class)
+                          eval_class, use_srn)
     logger.info('metric eval ***************')
     for k, v in metirc.items():
         logger.info('{}:{}'.format(k, v))