From 1c0e496530c7fe6ed2fb9bb9851b23983b41866b Mon Sep 17 00:00:00 2001 From: zhoujun Date: Tue, 13 Sep 2022 10:10:47 +0800 Subject: [PATCH] cp of 7550 (#7566) * fix table cpp infer bug * update vis * update doc * change default tale dict to ch --- deploy/cpp_infer/include/args.h | 1 + deploy/cpp_infer/include/paddlestructure.h | 13 +-- deploy/cpp_infer/include/postprocess_op.h | 15 ++- deploy/cpp_infer/include/structure_table.h | 7 +- deploy/cpp_infer/include/utility.h | 8 ++ deploy/cpp_infer/readme.md | 1 + deploy/cpp_infer/readme_ch.md | 1 + deploy/cpp_infer/src/args.cpp | 4 +- deploy/cpp_infer/src/main.cpp | 6 ++ deploy/cpp_infer/src/paddlestructure.cpp | 101 ++++++++++----------- deploy/cpp_infer/src/postprocess_op.cpp | 35 +++++-- deploy/cpp_infer/src/structure_table.cpp | 5 +- deploy/cpp_infer/src/utility.cpp | 74 ++++++++++++++- ppstructure/utility.py | 2 +- 14 files changed, 180 insertions(+), 93 deletions(-) diff --git a/deploy/cpp_infer/include/args.h b/deploy/cpp_infer/include/args.h index e0dd8bbc..f7fac9c9 100644 --- a/deploy/cpp_infer/include/args.h +++ b/deploy/cpp_infer/include/args.h @@ -54,6 +54,7 @@ DECLARE_string(table_model_dir); DECLARE_int32(table_max_len); DECLARE_int32(table_batch_num); DECLARE_string(table_char_dict_path); +DECLARE_bool(merge_no_span_structure); // forward related DECLARE_bool(det); DECLARE_bool(rec); diff --git a/deploy/cpp_infer/include/paddlestructure.h b/deploy/cpp_infer/include/paddlestructure.h index b30ac045..6d2c8b7d 100644 --- a/deploy/cpp_infer/include/paddlestructure.h +++ b/deploy/cpp_infer/include/paddlestructure.h @@ -54,15 +54,12 @@ private: std::vector &time_info_det, std::vector &time_info_rec, std::vector &time_info_cls); - std::string - rebuild_table(std::vector rec_html_tags, - std::vector>> rec_boxes, - std::vector &ocr_result); + std::string rebuild_table(std::vector rec_html_tags, + std::vector> rec_boxes, + std::vector &ocr_result); - float iou(std::vector> &box1, - std::vector> &box2); - float dis(std::vector> &box1, - std::vector> &box2); + float iou(std::vector &box1, std::vector &box2); + float dis(std::vector &box1, std::vector &box2); static bool comparison_dis(const std::vector &dis1, const std::vector &dis2) { diff --git a/deploy/cpp_infer/include/postprocess_op.h b/deploy/cpp_infer/include/postprocess_op.h index 77b3f8b6..f5db52a6 100644 --- a/deploy/cpp_infer/include/postprocess_op.h +++ b/deploy/cpp_infer/include/postprocess_op.h @@ -92,14 +92,13 @@ private: class TablePostProcessor { public: - void init(std::string label_path); - void - Run(std::vector &loc_preds, std::vector &structure_probs, - std::vector &rec_scores, std::vector &loc_preds_shape, - std::vector &structure_probs_shape, - std::vector> &rec_html_tag_batch, - std::vector>>> &rec_boxes_batch, - std::vector &width_list, std::vector &height_list); + void init(std::string label_path, bool merge_no_span_structure = true); + void Run(std::vector &loc_preds, std::vector &structure_probs, + std::vector &rec_scores, std::vector &loc_preds_shape, + std::vector &structure_probs_shape, + std::vector> &rec_html_tag_batch, + std::vector>> &rec_boxes_batch, + std::vector &width_list, std::vector &height_list); private: std::vector label_list_; diff --git a/deploy/cpp_infer/include/structure_table.h b/deploy/cpp_infer/include/structure_table.h index 7449c6cd..c09e6565 100644 --- a/deploy/cpp_infer/include/structure_table.h +++ b/deploy/cpp_infer/include/structure_table.h @@ -44,7 +44,8 @@ public: const int &gpu_mem, const int &cpu_math_library_num_threads, const bool &use_mkldnn, const string &label_path, const bool &use_tensorrt, const std::string &precision, - const int &table_batch_num, const int &table_max_len) { + const int &table_batch_num, const int &table_max_len, + const bool &merge_no_span_structure) { this->use_gpu_ = use_gpu; this->gpu_id_ = gpu_id; this->gpu_mem_ = gpu_mem; @@ -55,7 +56,7 @@ public: this->table_batch_num_ = table_batch_num; this->table_max_len_ = table_max_len; - this->post_processor_.init(label_path); + this->post_processor_.init(label_path, merge_no_span_structure); LoadModel(model_dir); } @@ -65,7 +66,7 @@ public: void Run(std::vector img_list, std::vector> &rec_html_tags, std::vector &rec_scores, - std::vector>>> &rec_boxes, + std::vector>> &rec_boxes, std::vector ×); private: diff --git a/deploy/cpp_infer/include/utility.h b/deploy/cpp_infer/include/utility.h index 520804f6..85b280fe 100644 --- a/deploy/cpp_infer/include/utility.h +++ b/deploy/cpp_infer/include/utility.h @@ -42,6 +42,7 @@ struct OCRPredictResult { struct StructurePredictResult { std::vector box; + std::vector> cell_box; std::string type; std::vector text_res; std::string html; @@ -56,6 +57,10 @@ public: const std::vector &ocr_result, const std::string &save_path); + static void VisualizeBboxes(const cv::Mat &srcimg, + const StructurePredictResult &structure_result, + const std::string &save_path); + template inline static size_t argmax(ForwardIterator first, ForwardIterator last) { return std::distance(first, std::max_element(first, last)); @@ -81,6 +86,9 @@ public: static void sorted_boxes(std::vector &ocr_result); + static std::vector xyxyxyxy2xyxy(std::vector> &box); + static std::vector xyxyxyxy2xyxy(std::vector &box); + private: static bool comparison_box(const OCRPredictResult &result1, const OCRPredictResult &result2) { diff --git a/deploy/cpp_infer/readme.md b/deploy/cpp_infer/readme.md index 2afdf795..2974f322 100644 --- a/deploy/cpp_infer/readme.md +++ b/deploy/cpp_infer/readme.md @@ -350,6 +350,7 @@ More parameters are as follows, |table_model_dir|string|-|Address of table recognition inference model| |table_char_dict_path|string|../../ppocr/utils/dict/table_structure_dict.txt|dictionary file| |table_max_len|int|488|The size of the long side of the input image of the table recognition model, the final input image size of the network is（table_max_len，table_max_len）| +|merge_no_span_structure|bool|true|Whether to merge and to 和 为| * PaddleOCR也支持多语言的预测，更多支持的语言和模型可以参考[识别文档](../../doc/doc_ch/recognition.md)中的多语言字典与模型部分，如果希望进行多语言预测，只需将修改`rec_char_dict_path`（字典文件路径）以及`rec_model_dir`（inference模型路径）字段即可。 diff --git a/deploy/cpp_infer/src/args.cpp b/deploy/cpp_infer/src/args.cpp index df1b9e32..17e9c8b6 100644 --- a/deploy/cpp_infer/src/args.cpp +++ b/deploy/cpp_infer/src/args.cpp @@ -55,8 +55,10 @@ DEFINE_int32(rec_img_w, 320, "rec image width"); DEFINE_string(table_model_dir, "", "Path of table struture inference model."); DEFINE_int32(table_max_len, 488, "max len size of input image."); DEFINE_int32(table_batch_num, 1, "table_batch_num."); +DEFINE_bool(merge_no_span_structure, true, + "Whether merge and to "); DEFINE_string(table_char_dict_path, - "../../ppocr/utils/dict/table_structure_dict.txt", + "../../ppocr/utils/dict/table_structure_dict_ch.txt", "Path of dictionary."); // ocr forward related diff --git a/deploy/cpp_infer/src/main.cpp b/deploy/cpp_infer/src/main.cpp index 66412a7b..34ffdc62 100644 --- a/deploy/cpp_infer/src/main.cpp +++ b/deploy/cpp_infer/src/main.cpp @@ -120,6 +120,7 @@ void structure(std::vector &cv_all_img_names) { engine.structure(cv_all_img_names, false, FLAGS_table); for (int i = 0; i < cv_all_img_names.size(); i++) { cout << "predict img: " << cv_all_img_names[i] << endl; + cv::Mat srcimg = cv::imread(cv_all_img_names[i], cv::IMREAD_COLOR); for (int j = 0; j < structure_results[i].size(); j++) { std::cout << j << "\ttype: " << structure_results[i][j].type << ", region: ["; @@ -129,6 +130,11 @@ void structure(std::vector &cv_all_img_names) { << structure_results[i][j].box[3] << "], res: "; if (structure_results[i][j].type == "table") { std::cout << structure_results[i][j].html << std::endl; + std::string file_name = Utility::basename(cv_all_img_names[i]); + + Utility::VisualizeBboxes(srcimg, structure_results[i][j], + FLAGS_output + "/" + std::to_string(j) + "_" + + file_name); } else { Utility::print_result(structure_results[i][j].text_res); } diff --git a/deploy/cpp_infer/src/paddlestructure.cpp b/deploy/cpp_infer/src/paddlestructure.cpp index 1ca85a96..ea69977a 100644 --- a/deploy/cpp_infer/src/paddlestructure.cpp +++ b/deploy/cpp_infer/src/paddlestructure.cpp @@ -27,7 +27,7 @@ PaddleStructure::PaddleStructure() { FLAGS_table_model_dir, FLAGS_use_gpu, FLAGS_gpu_id, FLAGS_gpu_mem, FLAGS_cpu_threads, FLAGS_enable_mkldnn, FLAGS_table_char_dict_path, FLAGS_use_tensorrt, FLAGS_precision, FLAGS_table_batch_num, - FLAGS_table_max_len); + FLAGS_table_max_len, FLAGS_merge_no_span_structure); } }; @@ -42,7 +42,7 @@ PaddleStructure::structure(std::vector cv_all_img_names, std::vector> structure_results; if (!Utility::PathExists(FLAGS_output) && FLAGS_det) { - mkdir(FLAGS_output.c_str(), 0777); + Utility::CreateDir(FLAGS_output); } for (int i = 0; i < cv_all_img_names.size(); ++i) { std::vector structure_result; @@ -84,7 +84,7 @@ void PaddleStructure::table(cv::Mat img, // predict structure std::vector> structure_html_tags; std::vector structure_scores(1, 0); - std::vector>>> structure_boxes; + std::vector>> structure_boxes; std::vector structure_imes; std::vector img_list; img_list.push_back(img); @@ -103,20 +103,15 @@ void PaddleStructure::table(cv::Mat img, this->det(img_list[i], ocr_result, time_info_det); // crop image std::vector rec_img_list; + std::vector ocr_box; for (int j = 0; j < ocr_result.size(); j++) { - int x_collect[4] = {ocr_result[j].box[0][0], ocr_result[j].box[1][0], - ocr_result[j].box[2][0], ocr_result[j].box[3][0]}; - int y_collect[4] = {ocr_result[j].box[0][1], ocr_result[j].box[1][1], - ocr_result[j].box[2][1], ocr_result[j].box[3][1]}; - int left = int(*std::min_element(x_collect, x_collect + 4)); - int right = int(*std::max_element(x_collect, x_collect + 4)); - int top = int(*std::min_element(y_collect, y_collect + 4)); - int bottom = int(*std::max_element(y_collect, y_collect + 4)); - std::vector box{max(0, left - expand_pixel), - max(0, top - expand_pixel), - min(img_list[i].cols, right + expand_pixel), - min(img_list[i].rows, bottom + expand_pixel)}; - cv::Mat crop_img = Utility::crop_image(img_list[i], box); + ocr_box = Utility::xyxyxyxy2xyxy(ocr_result[j].box); + ocr_box[0] = max(0, ocr_box[0] - expand_pixel); + ocr_box[1] = max(0, ocr_box[1] - expand_pixel), + ocr_box[2] = min(img_list[i].cols, ocr_box[2] + expand_pixel); + ocr_box[3] = min(img_list[i].rows, ocr_box[3] + expand_pixel); + + cv::Mat crop_img = Utility::crop_image(img_list[i], ocr_box); rec_img_list.push_back(crop_img); } // rec @@ -125,38 +120,37 @@ void PaddleStructure::table(cv::Mat img, html = this->rebuild_table(structure_html_tags[i], structure_boxes[i], ocr_result); structure_result.html = html; + structure_result.cell_box = structure_boxes[i]; structure_result.html_score = structure_scores[i]; } }; -std::string PaddleStructure::rebuild_table( - std::vector structure_html_tags, - std::vector>> structure_boxes, - std::vector &ocr_result) { +std::string +PaddleStructure::rebuild_table(std::vector structure_html_tags, + std::vector> structure_boxes, + std::vector &ocr_result) { // match text in same cell std::vector> matched(structure_boxes.size(), std::vector()); + std::vector ocr_box; + std::vector structure_box; for (int i = 0; i < ocr_result.size(); i++) { + ocr_box = Utility::xyxyxyxy2xyxy(ocr_result[i].box); + ocr_box[0] -= 1; + ocr_box[1] -= 1; + ocr_box[2] += 1; + ocr_box[3] += 1; std::vector> dis_list(structure_boxes.size(), std::vector(3, 100000.0)); for (int j = 0; j < structure_boxes.size(); j++) { - int x_collect[4] = {ocr_result[i].box[0][0], ocr_result[i].box[1][0], - ocr_result[i].box[2][0], ocr_result[i].box[3][0]}; - int y_collect[4] = {ocr_result[i].box[0][1], ocr_result[i].box[1][1], - ocr_result[i].box[2][1], ocr_result[i].box[3][1]}; - int left = int(*std::min_element(x_collect, x_collect + 4)); - int right = int(*std::max_element(x_collect, x_collect + 4)); - int top = int(*std::min_element(y_collect, y_collect + 4)); - int bottom = int(*std::max_element(y_collect, y_collect + 4)); - std::vector> box(2, std::vector(2, 0)); - box[0][0] = left - 1; - box[0][1] = top - 1; - box[1][0] = right + 1; - box[1][1] = bottom + 1; - - dis_list[j][0] = this->dis(box, structure_boxes[j]); - dis_list[j][1] = 1 - this->iou(box, structure_boxes[j]); + if (structure_boxes[i].size() == 8) { + structure_box = Utility::xyxyxyxy2xyxy(structure_boxes[j]); + } else { + structure_box = structure_boxes[j]; + } + dis_list[j][0] = this->dis(ocr_box, structure_box); + dis_list[j][1] = 1 - this->iou(ocr_box, structure_box); dis_list[j][2] = j; } // find min dis idx @@ -164,6 +158,7 @@ std::string PaddleStructure::rebuild_table( PaddleStructure::comparison_dis); matched[dis_list[0][2]].push_back(ocr_result[i].text); } + // get pred html std::string html_str = ""; int td_tag_idx = 0; @@ -221,19 +216,18 @@ std::string PaddleStructure::rebuild_table( return html_str; } -float PaddleStructure::iou(std::vector> &box1, - std::vector> &box2) { - int area1 = max(0, box1[1][0] - box1[0][0]) * max(0, box1[1][1] - box1[0][1]); - int area2 = max(0, box2[1][0] - box2[0][0]) * max(0, box2[1][1] - box2[0][1]); +float PaddleStructure::iou(std::vector &box1, std::vector &box2) { + int area1 = max(0, box1[2] - box1[0]) * max(0, box1[3] - box1[1]); + int area2 = max(0, box2[2] - box2[0]) * max(0, box2[3] - box2[1]); // computing the sum_area int sum_area = area1 + area2; // find the each point of intersect rectangle - int x1 = max(box1[0][0], box2[0][0]); - int y1 = max(box1[0][1], box2[0][1]); - int x2 = min(box1[1][0], box2[1][0]); - int y2 = min(box1[1][1], box2[1][1]); + int x1 = max(box1[0], box2[0]); + int y1 = max(box1[1], box2[1]); + int x2 = min(box1[2], box2[2]); + int y2 = min(box1[3], box2[3]); // judge if there is an intersect if (y1 >= y2 || x1 >= x2) { @@ -244,17 +238,16 @@ float PaddleStructure::iou(std::vector> &box1, } } -float PaddleStructure::dis(std::vector> &box1, - std::vector> &box2) { - int x1_1 = box1[0][0]; - int y1_1 = box1[0][1]; - int x2_1 = box1[1][0]; - int y2_1 = box1[1][1]; +float PaddleStructure::dis(std::vector &box1, std::vector &box2) { + int x1_1 = box1[0]; + int y1_1 = box1[1]; + int x2_1 = box1[2]; + int y2_1 = box1[3]; - int x1_2 = box2[0][0]; - int y1_2 = box2[0][1]; - int x2_2 = box2[1][0]; - int y2_2 = box2[1][1]; + int x1_2 = box2[0]; + int y1_2 = box2[1]; + int x2_2 = box2[2]; + int y2_2 = box2[3]; float dis = abs(x1_2 - x1_1) + abs(y1_2 - y1_1) + abs(x2_2 - x2_1) + abs(y2_2 - y2_1); diff --git a/deploy/cpp_infer/src/postprocess_op.cpp b/deploy/cpp_infer/src/postprocess_op.cpp index 551f98a1..4b0c693c 100644 --- a/deploy/cpp_infer/src/postprocess_op.cpp +++ b/deploy/cpp_infer/src/postprocess_op.cpp @@ -352,8 +352,21 @@ std::vector>> DBPostProcessor::FilterTagDetRes( return root_points; } -void TablePostProcessor::init(std::string label_path) { +void TablePostProcessor::init(std::string label_path, + bool merge_no_span_structure) { this->label_list_ = Utility::ReadDict(label_path); + if (merge_no_span_structure) { + this->label_list_.push_back(""); + std::vector::iterator it; + for (it = this->label_list_.begin(); it != this->label_list_.end();) { + if (*it == "") { + it = this->label_list_.erase(it); + } else { + ++it; + } + } + } + // add_special_char this->label_list_.insert(this->label_list_.begin(), this->beg); this->label_list_.push_back(this->end); } @@ -363,12 +376,12 @@ void TablePostProcessor::Run( std::vector &rec_scores, std::vector &loc_preds_shape, std::vector &structure_probs_shape, std::vector> &rec_html_tag_batch, - std::vector>>> &rec_boxes_batch, + std::vector>> &rec_boxes_batch, std::vector &width_list, std::vector &height_list) { for (int batch_idx = 0; batch_idx < structure_probs_shape[0]; batch_idx++) { // image tags and boxs std::vector rec_html_tags; - std::vector>> rec_boxes; + std::vector> rec_boxes; float score = 0.f; int count = 0; @@ -378,7 +391,7 @@ void TablePostProcessor::Run( // step for (int step_idx = 0; step_idx < structure_probs_shape[1]; step_idx++) { std::string html_tag; - std::vector> rec_box; + std::vector rec_box; // html tag int step_start_idx = (batch_idx * structure_probs_shape[1] + step_idx) * structure_probs_shape[2]; @@ -399,17 +412,19 @@ void TablePostProcessor::Run( count += 1; score += char_score; rec_html_tags.push_back(html_tag); + // box if (html_tag == "" || html_tag == "") { - for (int point_idx = 0; point_idx < loc_preds_shape[2]; - point_idx += 2) { - std::vector point(2, 0); + for (int point_idx = 0; point_idx < loc_preds_shape[2]; point_idx++) { step_start_idx = (batch_idx * structure_probs_shape[1] + step_idx) * loc_preds_shape[2] + point_idx; - point[0] = int(loc_preds[step_start_idx] * width_list[batch_idx]); - point[1] = - int(loc_preds[step_start_idx + 1] * height_list[batch_idx]); + float point = loc_preds[step_start_idx]; + if (point_idx % 2 == 0) { + point = int(point * width_list[batch_idx]); + } else { + point = int(point * height_list[batch_idx]); + } rec_box.push_back(point); } rec_boxes.push_back(rec_box); diff --git a/deploy/cpp_infer/src/structure_table.cpp b/deploy/cpp_infer/src/structure_table.cpp index bbc32580..7df0ab94 100644 --- a/deploy/cpp_infer/src/structure_table.cpp +++ b/deploy/cpp_infer/src/structure_table.cpp @@ -20,7 +20,7 @@ void StructureTableRecognizer::Run( std::vector img_list, std::vector> &structure_html_tags, std::vector &structure_scores, - std::vector>>> &structure_boxes, + std::vector>> &structure_boxes, std::vector ×) { std::chrono::duration preprocess_diff = std::chrono::steady_clock::now() - std::chrono::steady_clock::now(); @@ -89,8 +89,7 @@ void StructureTableRecognizer::Run( auto postprocess_start = std::chrono::steady_clock::now(); std::vector> structure_html_tag_batch; std::vector structure_score_batch; - std::vector>>> - structure_boxes_batch; + std::vector>> structure_boxes_batch; this->post_processor_.Run(loc_preds, structure_probs, structure_score_batch, predict_shape0, predict_shape1, structure_html_tag_batch, structure_boxes_batch, diff --git a/deploy/cpp_infer/src/utility.cpp b/deploy/cpp_infer/src/utility.cpp index 4bfc1d09..0e6ba17f 100644 --- a/deploy/cpp_infer/src/utility.cpp +++ b/deploy/cpp_infer/src/utility.cpp @@ -65,6 +65,37 @@ void Utility::VisualizeBboxes(const cv::Mat &srcimg, << std::endl; } +void Utility::VisualizeBboxes(const cv::Mat &srcimg, + const StructurePredictResult &structure_result, + const std::string &save_path) { + cv::Mat img_vis; + srcimg.copyTo(img_vis); + for (int n = 0; n < structure_result.cell_box.size(); n++) { + if (structure_result.cell_box[n].size() == 8) { + cv::Point rook_points[4]; + for (int m = 0; m < structure_result.cell_box[n].size(); m += 2) { + rook_points[m / 2] = + cv::Point(int(structure_result.cell_box[n][m]), + int(structure_result.cell_box[n][m + 1])); + } + const cv::Point *ppt[1] = {rook_points}; + int npt[] = {4}; + cv::polylines(img_vis, ppt, npt, 1, 1, CV_RGB(0, 255, 0), 2, 8, 0); + } else if (structure_result.cell_box[n].size() == 4) { + cv::Point rook_points[2]; + rook_points[0] = cv::Point(int(structure_result.cell_box[n][0]), + int(structure_result.cell_box[n][1])); + rook_points[1] = cv::Point(int(structure_result.cell_box[n][2]), + int(structure_result.cell_box[n][3])); + cv::rectangle(img_vis, rook_points[0], rook_points[1], CV_RGB(0, 255, 0), + 2, 8, 0); + } + } + + cv::imwrite(save_path, img_vis); + std::cout << "The table visualized image saved in " + save_path << std::endl; +} + // list all files under a directory void Utility::GetAllFiles(const char *dir_name, std::vector &all_inputs) { @@ -268,13 +299,46 @@ cv::Mat Utility::crop_image(cv::Mat &img, std::vector &area) { void Utility::sorted_boxes(std::vector &ocr_result) { std::sort(ocr_result.begin(), ocr_result.end(), Utility::comparison_box); - - for (int i = 0; i < ocr_result.size() - 1; i++) { - if (abs(ocr_result[i + 1].box[0][1] - ocr_result[i].box[0][1]) < 10 && - (ocr_result[i + 1].box[0][0] < ocr_result[i].box[0][0])) { - std::swap(ocr_result[i], ocr_result[i + 1]); + if (ocr_result.size() > 0) { + for (int i = 0; i < ocr_result.size() - 1; i++) { + for (int j = i; j > 0; j--) { + if (abs(ocr_result[j + 1].box[0][1] - ocr_result[j].box[0][1]) < 10 && + (ocr_result[j + 1].box[0][0] < ocr_result[j].box[0][0])) { + std::swap(ocr_result[i], ocr_result[i + 1]); + } + } } } } +std::vector Utility::xyxyxyxy2xyxy(std::vector> &box) { + int x_collect[4] = {box[0][0], box[1][0], box[2][0], box[3][0]}; + int y_collect[4] = {box[0][1], box[1][1], box[2][1], box[3][1]}; + int left = int(*std::min_element(x_collect, x_collect + 4)); + int right = int(*std::max_element(x_collect, x_collect + 4)); + int top = int(*std::min_element(y_collect, y_collect + 4)); + int bottom = int(*std::max_element(y_collect, y_collect + 4)); + std::vector box1(4, 0); + box1[0] = left; + box1[1] = top; + box1[2] = right; + box1[3] = bottom; + return box1; +} + +std::vector Utility::xyxyxyxy2xyxy(std::vector &box) { + int x_collect[4] = {box[0], box[2], box[4], box[6]}; + int y_collect[4] = {box[1], box[3], box[5], box[7]}; + int left = int(*std::min_element(x_collect, x_collect + 4)); + int right = int(*std::max_element(x_collect, x_collect + 4)); + int top = int(*std::min_element(y_collect, y_collect + 4)); + int bottom = int(*std::max_element(y_collect, y_collect + 4)); + std::vector box1(4, 0); + box1[0] = left; + box1[1] = top; + box1[2] = right; + box1[3] = bottom; + return box1; +} + } // namespace PaddleOCR \ No newline at end of file diff --git a/ppstructure/utility.py b/ppstructure/utility.py index bdea0af6..97b6d6fe 100644 --- a/ppstructure/utility.py +++ b/ppstructure/utility.py @@ -32,7 +32,7 @@ def init_args(): parser.add_argument( "--table_char_dict_path", type=str, - default="../ppocr/utils/dict/table_structure_dict.txt") + default="../ppocr/utils/dict/table_structure_dict_ch.txt") # params for layout parser.add_argument("--layout_model_dir", type=str) parser.add_argument( -- GitLab