Created by: mcl-stone

使用detection0.2 版本yolov3_darknet_voc训练出的模型，再使用了最新版本的detection0.3，按照paddleserving部署模型导出时报错。

/PaddleDetection-release-0.3# python3 tools/export_serving_model.py -c configs/dcn/yolov3_darknet_voc_diy.yml --output_dir=./inference_model -o weights=/home/mcl/Desktop/dir_bestmodel/yolov3_darknet_voc_diy/model_final
2020-05-19 20:59:41,330-INFO: The 'num_classes'(number of classes) you set is 1, and 'with_background' in 'dataset' sets False.
So please note the actual number of categories is 1.
2020-05-19 20:59:44,537-INFO: save_inference_model pruned unused feed variables im_id
2020-05-19 20:59:44,537-INFO: Export serving model to ./inference_model, client side: ./inference_model/serving_client, server side: ./inference_model/serving_server. input: ['image', 'im_size'], output: ['multiclass_nms_0.tmp_0']...
Traceback (most recent call last):
  File "tools/export_serving_model.py", line 97, in <module>
    main()
  File "tools/export_serving_model.py", line 84, in main
    save_serving_model(FLAGS, exe, feed_vars, test_fetches, infer_prog)
  File "tools/export_serving_model.py", line 54, in save_serving_model
    infer_prog)
  File "/usr/local/lib/python3.6/dist-packages/paddle_serving_client/io/__init__.py", line 42, in save_model
    main_program=main_program)
  File "/usr/local/lib/python3.6/dist-packages/paddle/fluid/io.py", line 1217, in save_inference_model
    feeded_var_names=feeded_var_names, targets=target_vars)
  File "/usr/local/lib/python3.6/dist-packages/paddle/fluid/framework.py", line 4125, in _prune_with_input
    raise ValueError("All targets of prune() can only be "
ValueError: All targets of prune() can only be Variable or Operator.

求帮助