1、出现该错误表示opencv冲突，该机器上有多个opencv版本。

解决方法：卸载低版本opencv

2、bazel中BUILD的写法：

1. copts中放置-I/usr/include/、-D
2. linkopts中放置-lopencv_core、-std=c++11、-L/usr/lib64
3. 出现编译错误时，可以使用--verbose_failures来打印错误处的详细编译命令，以此查找脚本的错误
   bazel build tensorflow/... --verbose_failures

    

4. 1 package( 2     default_visibility = ["//tensorflow_serving:internal"], 3     features = ["no_layering_check"], 4 ) 5  6 licenses(["notice"])  # Apache 2.0 7  8 exports_files(["LICENSE"]) 9 10 load("//tensorflow_serving:serving.bzl", "serving_proto_library")11 12 filegroup(13     name = "all_files",14     srcs = glob(15         ["**/*"],16         exclude = [17             "**/METADATA",18             "**/OWNERS",19         ],20     ),21 )22 23 cc_library(24     name = "recog_proc",25     srcs = glob([26         "ImageProc.cpp",27     ]),28     hdrs = [29         "ImageProc.h",30     ],31     copts = [32         "-I/usr/include/",33         "-I/usr/include/ImageMagick/",34     ],35     deps = [36         "//tensorflow_serving/apis:prediction_service_proto",37         "@grpc//:grpc++",38         "@org_tensorflow//tensorflow/core:framework",39         "@protobuf_archive//:protobuf_lite",40     ],41     visibility = [ 42         "//visibility:public",43     ],44 )45 46 cc_binary(47     name = "recog_srv_cc",48     srcs = glob([49         "main.cpp",50     ]),51     copts = [52         "-I/usr/include/",53         "-L/usr/lib64",54         "-DSS_LOG_DEFAULT_MODULE=recog",55     ],56     linkopts = [57         "-lACE",58         "-lssplatform",59         "-ljsoncpp",60         "-lopencv_core",61         "-lopencv_imgproc",62         "-lopencv_highgui",63         "-lopencv_imgcodecs",64         "-lopencv_photo",65         "-lMagick++",66         "-lMagickCore",67         "-lMagickWand",68         "-Wl,-no-undefined",69         "-Wno-reorder",70         "-std=c++11",71         "-pthread",72         "-g",73         "-L/usr/local/lib64",74     ],75     deps = [76         ":recog_proc",77     ],78 )

    

5. 示例：