Created by: OliverLPH

System information -PaddlePaddle version: develop, 028de85 -CPU: Intel(R) Xeon(R) CPU E5-2650 v4 @ 2.20GHz, mkldnn enabled -GPU: P4, CUDA10, CUDNN7.5, compile with GPU, but run cpu inference -OS Platform: Ubuntu1604 -Dockerfile: Paddle/tools/manylinux1/Dockerfile.cuda10_cudnn7_gcc48_ubuntu16 -Python version: 3.7 -Cmake orders

cmake .. -DPYTHON_EXECUTABLE:FILEPATH=/usr/local/bin/python3.7 \
         -DPYTHON_INCLUDE_DIR:PATH=/usr/local/include/python3.7m \
         -DPYTHON_LIBRARIES:FILEPATH=/usr/local/lib/libpython3.7m.so \
         -DWITH_FLUID_ONLY=ON \
         -DWITH_GPU=ON \
         -DWITH_TESTING=OFF \
         -DCMAKE_BUILD_TYPE=Release \
         -DON_INFER=ON \
         -DWITH_INFERENCE_API_TEST=OFF \
         -DWITH_MKL=ON \
         -DWITH_MKLDNN=ON \
         -DTENSORRT_ROOT=/usr

-C++version.txt -API information

config = AnalysisConfig(args.model_file, args.params_file)
config.disable_gpu()
config.enable_mkldnn()
config.set_cpu_math_library_num_threads(4)
config.switch_use_feed_fetch_ops(False)
config.switch_specify_input_names(True)
config.enable_memory_optim()

Note: You can get most of the information by running summary_env.py. To Reproduce Steps to reproduce the behavior

python3.7 infer.py --model_file=./mask_rcnn_r50_1x/model/__model__ \
                                --params_file=./mask_rcnn_r50_1x/model/__params__ \
                                --input_file=./mask_rcnn_r50_1x/data/image.txt

Describe your current behavior

----------------------
Error Message Summary:
----------------------
Error: The Input Variable(X) of roi_align Op used to determine kernel data type is empty or not LoDTensor or SelectedRows or LoDTensorArray.
  [Hint: Expected data_type != dafault_data_type, but received data_type:-1 == dafault_data_type:-1.] at (/workspace/paddle/fluid/framework/operator.cc:1339)
  [operator < roi_align > error]

Code to reproduce the issue infer.txt

Other info / logs