我正在为Tensorflow编写一个应该加载视频的自定义操作。为此,我需要包含OpenCV。
目前,该操作只是尝试打开 VideoCapture 并返回空张量。
这是C ++代码:
class Calculator:
def addition(self,x,y):
added = x + y
return added
def subtraction(self,x,y):
subtracted = x - y
return subtracted
def multiplication(self,x,y):
multiplied = x * y
return multiplied
def division(self,x,y):
divided = x / y
return divided
calculator = Calculator()
num1 = raw_input('First Number >')
num2 = raw_input('Second Number >')
print("1 \tAddition")
print("2 \tSubtraction")
print("3 \tMultiplication")
print("4 \tDivision")
operations = raw_input('Select operation number>')
if int(operations)== 1:
print (calculator.addition(float(num1),float(num2)))
if int(operations)== 2:
print (calculator.subtraction(float(num1),float(num2)))
if int(operations)== 3:
print (calculator.multiplication(float(num1),float(num2)))
if int(operations)== 4:
print (calculator.division(float(num1),float(num2)))
然后,我使用以下命令编译代码:
#include "opencv2/opencv.hpp"
#include "opencv2/core/core.hpp"
#include "opencv2/highgui/highgui.hpp"
#include "tensorflow/core/framework/op.h"
#include "tensorflow/core/framework/shape_inference.h"
#include "tensorflow/core/framework/op_kernel.h"
#include <iostream>
using namespace tensorflow;
using namespace cv;
using namespace std;
using shape_inference::ShapeHandle;
using shape_inference::DimensionHandle;
REGISTER_OP("LoadVideo")
.Input("filename: string")
.Output("frame: float32")
.SetShapeFn([](::tensorflow::shape_inference::InferenceContext* c) {
TensorShape outputTensorShape({224, 224, 3});
ShapeHandle outputShapeHandle;
c->MakeShapeFromTensorShape(outputTensorShape, &outputShapeHandle);
c->set_output(0, outputShapeHandle);
return Status::OK();
});
class LoadVideoOp : public OpKernel {
public:
explicit LoadVideoOp(OpKernelConstruction* context) : OpKernel(context) {}
void Compute(OpKernelContext* context) override {
// Grab the input tensor
const Tensor& input_tensor = context->input(0);
auto input = input_tensor.flat<string>();
string filename = input(0);
VideoCapture cap = VideoCapture("data/0eRkpTGq5pA.mp4");
Tensor* output_tensor = NULL;
OP_REQUIRES_OK(context, context->allocate_output(0, {224, 224, 3}, &output_tensor));
}
};
REGISTER_KERNEL_BUILDER(Name("LoadVideo").Device(DEVICE_CPU), LoadVideoOp);
当我将编译后的代码加载到Python脚本中(使用 tf.load_op_library )并尝试运行操作时,我收到以下错误:
tensorflow.python.framework.errors_impl.NotFoundError:lib / ops / load_video.so:undefined symbol:_ZN2cv12VideoCaptureC1ERKNSt7__cxx1112basic_stringIcSt11char_traitsIcESaIcEEE
看起来编译的C ++代码无法访问相应的OpenCV对象。我不太了解C ++编译和链接,所以问题可能是我以错误的方式编译自定义操作。
请您帮我编译op,以便成功加载并运行tensorflow?
编辑1:
这是我用来加载自定义操作的Python脚本:
g++ -std=c++11 -shared -fPIC \
-I /home/master/anaconda3/envs/tf/lib/python3.6/site-packages/tensorflow/include \
-I ~/anaconda3/envs/tf/include/opencv2/ -I ~/anaconda3/envs/tf/include/opencv/ -O2 \
-L ~/anaconda3/envs/tf/lib \
load_video.cc -o load_video.so \
-lopencv_core -lopencv_videoio -lopencv_highgui \
-lopencv_imgproc -lopencv_video -lopencv_objdetect
错误发生在第2行(即尝试加载已编译的C ++代码时)。
解决方案:
我在重建OpenCV后成功编译并运行了自定义tensorflow操作。编译命令是:
import tensorflow as tf
load_video_module = tf.load_op_library('lib/ops/load_video.so')
with tf.Session():
x = load_video_module.load_video("data/0eRkpTGq5pA.mp4").eval()
print(x)
答案 0 :(得分:2)
您可以使用ldd
检查您的图书馆是否需要丢失库。
只需检查ldd load_video.so
。
但是,您可能并未将某些共享库与某些您正在使用的OpenCV方法相关联。
要确保链接并包含所需的每个库,您可以使用pkg-config
。
删除指向OpenCV libs的手动-I
和-l
标记,然后添加执行完成工作的pkg-config --libs --cflags opencv
(包含和链接库)为你
答案 1 :(得分:1)
如果您在Linux上运行,请进入包含视频文件的目录,然后执行,不带括号:
sudo chmod 777 (the name of your video file)
这应该让您的程序可以访问视频文件。 我不太了解C ++,但TensorFlow在被拒绝许可时经常会抛出此错误,所以请试一试并祝你好运!