Files
FastDeploy/fastdeploy/vision/ocr/ppocr/det_preprocessor.h
Wang Xinyu 91a1c72f98 [CVCUDA] PP-OCR detector preprocessor integrate CV-CUDA (#1382)
* move manager initialized_ flag to ppcls

* update dbdetector preprocess api

* declare processor op

* ppocr detector preprocessor support cvcuda

* move cvcuda op to class member

* ppcls use manager register api

* refactor det preprocessor init api

* add set preprocessor api

* add create processor macro

* new processor call api

* ppcls preprocessor init resize on cpu

* ppocr detector preprocessor set normalize api

* revert ppcls pybind

* remove dbdetector set preprocessor

* refine dbdetector preprocessor includes

* remove mean std in py constructor

* add comments

* update comment

* Update __init__.py
2023-02-22 19:39:11 +08:00

75 lines
2.8 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 "fastdeploy/vision/common/processors/manager.h"
#include "fastdeploy/vision/common/processors/resize.h"
#include "fastdeploy/vision/common/processors/pad.h"
#include "fastdeploy/vision/common/processors/normalize_and_permute.h"
#include "fastdeploy/vision/common/result.h"
namespace fastdeploy {
namespace vision {
namespace ocr {
/*! @brief Preprocessor object for DBDetector serials model.
*/
class FASTDEPLOY_DECL DBDetectorPreprocessor : public ProcessorManager {
public:
DBDetectorPreprocessor();
/** \brief Process the input image and prepare input tensors for runtime
*
* \param[in] image_batch The input image batch
* \param[in] outputs The output tensors which will feed in runtime
* \return true if the preprocess successed, otherwise false
*/
virtual bool Apply(FDMatBatch* image_batch, std::vector<FDTensor>* outputs);
/// Set max_side_len for the detection preprocess, default is 960
void SetMaxSideLen(int max_side_len) { max_side_len_ = max_side_len; }
/// Get max_side_len of the detection preprocess
int GetMaxSideLen() const { return max_side_len_; }
/// Set preprocess normalize parameters, please call this API to customize
/// the normalize parameters, otherwise it will use the default normalize
/// parameters.
void SetNormalize(const std::vector<float>& mean = {0.485f, 0.456f, 0.406f},
const std::vector<float>& std = {0.229f, 0.224f, 0.225f},
bool is_scale = true) {
normalize_permute_op_ =
std::make_shared<NormalizeAndPermute>(mean, std, is_scale);
}
/// Get the image info of the last batch, return a list of array
/// {image width, image height, resize width, resize height}
const std::vector<std::array<int, 4>>* GetBatchImgInfo() {
return &batch_det_img_info_;
}
private:
bool ResizeImage(FDMat* img, int resize_w, int resize_h, int max_resize_w,
int max_resize_h);
int max_side_len_ = 960;
std::vector<std::array<int, 4>> batch_det_img_info_;
std::shared_ptr<Resize> resize_op_;
std::shared_ptr<Pad> pad_op_;
std::shared_ptr<NormalizeAndPermute> normalize_permute_op_;
};
} // namespace ocr
} // namespace vision
} // namespace fastdeploy