Files
FastDeploy/fastdeploy/backends/openvino/ov_backend.h
heliqi 42f1888bb0 [Serving][Backend] Backend support zero_copy_infer and Serving reduce the output memory copy (#703)
* backend add zero copy infer interface

* fix bug

* fix bug

* fix bug

* paddle ipu
2022-11-28 14:07:53 +08:00

78 lines
2.4 KiB
C++

// Copyright (c) 2022 PaddlePaddle Authors. All Rights Reserved.
//
// Licensed under the Apache License, Version 2.0 (the "License");
// you may not use this file except in compliance with the License.
// You may obtain a copy of the License at
//
// http://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the License is distributed on an "AS IS" BASIS,
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// See the License for the specific language governing permissions and
// limitations under the License.
#pragma once
#include <iostream>
#include <memory>
#include <string>
#include <vector>
#include "fastdeploy/backends/backend.h"
#include "fastdeploy/utils/unique_ptr.h"
#include "openvino/openvino.hpp"
namespace fastdeploy {
struct OpenVINOBackendOption {
std::string device = "CPU";
int cpu_thread_num = -1;
int num_streams = 0;
std::map<std::string, std::vector<int64_t>> shape_infos;
std::set<std::string> cpu_operators{"MulticlassNms"};
};
class OpenVINOBackend : public BaseBackend {
public:
static ov::Core core_;
OpenVINOBackend() {}
virtual ~OpenVINOBackend() = default;
bool
InitFromPaddle(const std::string& model_file, const std::string& params_file,
const OpenVINOBackendOption& option = OpenVINOBackendOption());
bool
InitFromOnnx(const std::string& model_file,
const OpenVINOBackendOption& option = OpenVINOBackendOption());
bool Infer(std::vector<FDTensor>& inputs,
std::vector<FDTensor>* outputs,
bool copy_to_fd = true) override;
int NumInputs() const override;
int NumOutputs() const override;
TensorInfo GetInputInfo(int index) override;
TensorInfo GetOutputInfo(int index) override;
std::vector<TensorInfo> GetInputInfos() override;
std::vector<TensorInfo> GetOutputInfos() override;
std::unique_ptr<BaseBackend> Clone(void* stream = nullptr,
int device_id = -1) override;
private:
void InitTensorInfo(const std::vector<ov::Output<ov::Node>>& ov_outputs,
std::map<std::string, TensorInfo>* tensor_infos);
ov::CompiledModel compiled_model_;
ov::InferRequest request_;
OpenVINOBackendOption option_;
std::vector<TensorInfo> input_infos_;
std::vector<TensorInfo> output_infos_;
};
} // namespace fastdeploy