add layout and block model (#3967)

docs/module_usage/tutorials/ocr_modules/layout_detection.en.md

@@ -9,6 +9,31 @@ The core task of structure analysis is to parse and segment the content of input
 
 ## II. Supported Model List
 
+* <b>The layout detection model includes 20 common categories: document title, paragraph title, text, page number, abstract, table, references, footnotes, header, footer, algorithm, formula, formula number, image, table, seal, figure_table title, chart, and sidebar text and lists of references</b>
+<table>
+<thead>
+<tr>
+<th>Model</th><th>Model Download Link</th>
+<th>mAP(0.5) (%)</th>
+<th>GPU Inference Time (ms)<br/>[Normal Mode / High-Performance Mode]</th>
+<th>CPU Inference Time (ms)<br/>[Normal Mode / High-Performance Mode]</th>
+<th>Model Storage Size (M)</th>
+<th>Introduction</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+<td>PP-DocLayout_plus-L</td><td><a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_inference_model/paddle3.0.0/PP-DocLayout_plus-L_infer.tar">Inference Model</a>/<a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_pretrained_model/PP-DocLayout_plus-L_pretrained.pdparams">Training Model</a></td>
+<td>83.2</td>
+<td>34.6244 / 10.3945</td>
+<td>510.57 / - </td>
+<td>126.01 M</td>
+<td>A higher-precision layout area localization model trained on a self-built dataset containing Chinese and English papers, PPT, multi-layout magazines, contracts, books, exams, ancient books and research reports using RT-DETR-L</td>
+</tr>
+<tr>
+</tbody>
+</table>
+
 * <b>The layout detection model includes 23 common categories: document title, paragraph title, text, page number, abstract, table of contents, references, footnotes, header, footer, algorithm, formula, formula number, image, figure caption, table, table caption, seal, figure title, figure, header image, footer image, and sidebar text</b>
 <table>
 <thead>
@@ -49,7 +74,7 @@ The core task of structure analysis is to parse and segment the content of input
 </tbody>
 </table>
 
-> ❗ The above list includes the <b>3 core models</b> that are key supported by the text recognition module. The module actually supports a total of <b>11 full models</b>, including several predefined models with different categories. The complete model list is as follows:
+> ❗ The above list includes the <b>4 core models</b> that are key supported by the text recognition module. The module actually supports a total of <b>12 full models</b>, including several predefined models with different categories. The complete model list is as follows:
 
 <details><summary> 👉 Details of Model List</summary>
 

docs/module_usage/tutorials/ocr_modules/layout_detection.md

@@ -9,6 +9,33 @@ comments: true
 
 ## 二、支持模型列表
 
+* <b>版面检测模型，包含20个常见的类别：文档标题、段落标题、文本、页码、摘要、目录、参考文献、脚注、页眉、页脚、算法、公式、公式编号、图像、表格、图和表标题（图标题、表格标题和图表标题）、印章、图表、侧栏文本和参考文献内容</b>
+<table>
+<thead>
+<tr>
+<th>模型</th><th>模型下载链接</th>
+<th>mAP(0.5)（%）</th>
+<th>GPU推理耗时（ms）<br/>[常规模式 / 高性能模式]</th>
+<th>CPU推理耗时（ms）<br/>[常规模式 / 高性能模式]</th>
+<th>模型存储大小（M）</th>
+<th>介绍</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+<td>PP-DocLayout_plus-L</td><td><a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_inference_model/paddle3.0.0/PP-DocLayout_plus-L_infer.tar">推理模型</a>/<a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_pretrained_model/PP-DocLayout_plus-L_pretrained.pdparams">训练模型</a></td>
+<td>83.2</td>
+<td>34.6244 / 10.3945</td>
+<td>510.57 / - </td>
+<td>126.01 M</td>
+<td>基于RT-DETR-L在包含中英文论文、多栏杂志、报纸、PPT、合同、书本、试卷、研报、古籍、日文文档、竖版文字文档等场景的自建数据集训练的更高精度版面区域定位模型</td>
+</tr>
+<tr>
+</tbody>
+</table>
+
+<b>注：以上精度指标的评估集是自建的版面区域检测数据集，包含中英文论文、杂志、报纸、研报、PPT、试卷、课本等 1300 张文档类型图片。</b>
+
 * <b>版面检测模型，包含23个常见的类别：文档标题、段落标题、文本、页码、摘要、目录、参考文献、脚注、页眉、页脚、算法、公式、公式编号、图像、图表标题、表格、表格标题、印章、图表标题、图表、页眉图像、页脚图像、侧栏文本</b>
 <table>
 <thead>
@@ -50,7 +77,7 @@ comments: true
 </table>
 
 
-> ❗ 以上列出的是版面检测模块重点支持的<b>3个核心模型</b>，该模块总共支持<b>11个全量模型</b>，包含多个预定义了不同类别的模型，完整的模型列表如下：
+> ❗ 以上列出的是版面检测模块重点支持的<b>4个核心模型</b>，该模块总共支持<b>12个全量模型</b>，包含多个预定义了不同类别的模型，完整的模型列表如下：
 
 <details><summary> 👉模型列表详情</summary>
 
@@ -186,7 +213,8 @@ comments: true
           <ul>
               <li><strong>测试数据集：</strong>
                  <ul>
-                    <li>版面检测模型： PaddleOCR 自建的版面区域检测数据集，包含中英文论文、杂志、合同、书本、试卷和研报等常见的 500 张文档类型图片。</li>
+                    <li>20类版面检测模型： PaddleOCR 自建的版面区域检测数据集，包含中英文论文、杂志、报纸、研报、PPT、试卷、课本等 1300 张文档类型图片。</li>
+                    <li>23类版面检测模型： PaddleOCR 自建的版面区域检测数据集，包含中英文论文、杂志、合同、书本、试卷和研报等常见的 500 张文档类型图片。</li>
                     <li>表格版面检测模型：PaddleOCR 自建的版面表格区域检测数据集，包含中英文 7835 张带有表格的论文文档类型图片。</li>
                     <li>3类版面检测模型：PaddleOCR 自建的版面区域检测数据集，包含中英文论文、杂志和研报等常见的 1154 张文档类型图片。</li>
                     <li>5类英文文档区域检测模型： <a href="https://developer.ibm.com/exchanges/data/all/publaynet" target="_blank">PubLayNet</a> 的评估数据集，包含英文文档的 11245 张图片。</li>

docs/support_list/models_list.md

@@ -2341,6 +2341,80 @@ devanagari_PP-OCRv3_mobile_rec_infer.tar">推理模型</a>/<a href="">训练模
 
 ## [版面区域检测模块](../module_usage/tutorials/ocr_modules/layout_detection.md)
 
+* <b>版面检测模型，包含20个常见的类别：文档标题、段落标题、文本、页码、摘要、目录、参考文献、脚注、页眉、页脚、算法、公式、公式编号、图像、表格、图和表标题（图标题、表格标题和图表标题）、印章、图表、侧栏文本和参考文献内容</b>
+<table>
+<thead>
+<tr>
+<th>模型</th>
+<th>mAP(0.5)（%）</th>
+<th>GPU推理耗时（ms）<br/>[常规模式 / 高性能模式]</th>
+<th>CPU推理耗时（ms）<br/>[常规模式 / 高性能模式]</th>
+<th>模型存储大小（M）</th>
+<th>yaml文件</th>
+<th>模型下载链接</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+<td>PP-DocLayout_plus-L</td>
+<td>83.2</td>
+<td>34.6244 / 10.3945</td>
+<td>510.57 / - </td>
+<td>126.01 </td>
+<td><a href="https://github.com/PaddlePaddle/PaddleX/blob/develop/paddlex/configs/modules/layout_detection/PP-DocLayout_plus-L.yaml">PP-DocLayout_plus-L.yaml</a></td>
+<td><a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_inference_model/paddle3.0.0/PP-DocLayout_plus-L_infer.tar">推理模型</a>/<a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_pretrained_model/PP-DocLayout_plus-L_pretrained.pdparams">训练模型</a></td>
+</tr>
+</tbody>
+</table>
+
+<b>注：以上精度指标的评估集是自建的版面区域检测数据集，包含中英文论文、杂志、报纸、研报、PPT、试卷、课本等 1300 张文档类型图片。</b>
+
+* <b>版面检测模型，包含23个常见的类别：文档标题、段落标题、文本、页码、摘要、目录、参考文献、脚注、页眉、页脚、算法、公式、公式编号、图像、图表标题、表格、表格标题、印章、图表标题、图表、页眉图像、页脚图像、侧栏文本</b>
+<table>
+<thead>
+<tr>
+<th>模型</th>
+<th>mAP(0.5)（%）</th>
+<th>GPU推理耗时（ms）<br/>[常规模式 / 高性能模式]</th>
+<th>CPU推理耗时（ms）<br/>[常规模式 / 高性能模式]</th>
+<th>模型存储大小（M）</th>
+<th>yaml文件</th>
+<th>模型下载链接</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+<td>PP-DocLayout-L</td>
+<td>90.4</td>
+<td>34.6244 / 10.3945</td>
+<td>510.57 / - </td>
+<td>123.76 </td>
+<td><a href="https://github.com/PaddlePaddle/PaddleX/blob/develop/paddlex/configs/modules/layout_detection/PP-DocLayout-L.yaml">PP-DocLayout-L.yaml</a></td>
+<td><a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_inference_model/paddle3.0.0/PP-DocLayout-L_infer.tar">推理模型</a>/<a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_pretrained_model/PP-DocLayout-L_pretrained.pdparams">训练模型</a></td>
+</tr>
+<tr>
+<td>PP-DocLayout-M</td>
+<td>75.2</td>
+<td>13.3259 / 4.8685</td>
+<td>44.0680 / 44.0680</td>
+<td>22.578</td>
+<td><a href="https://github.com/PaddlePaddle/PaddleX/blob/develop/paddlex/configs/modules/layout_detection/PP-DocLayout-M.yaml">PP-DocLayout-M.yaml</a></td>
+<td><a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_inference_model/paddle3.0.0/PP-DocLayout-M_infer.tar">推理模型</a>/<a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_pretrained_model/PP-DocLayout-M_pretrained.pdparams">训练模型</a></td>
+</tr>
+<tr>
+<td>PP-DocLayout-S</td>
+<td>70.9</td>
+<td>8.3008 / 2.3794</td>
+<td>10.0623 / 9.9296</td>
+<td>4.834</td>
+<td><a href="https://github.com/PaddlePaddle/PaddleX/blob/develop/paddlex/configs/modules/layout_detection/PP-DocLayout-S.yaml">PP-DocLayout-S.yaml</a></td>
+<td><a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_inference_model/paddle3.0.0/PP-DocLayout-S_infer.tar">推理模型</a>/<a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_pretrained_model/PP-DocLayout-S_pretrained.pdparams">训练模型</a></td>
+</tr>
+</tbody>
+</table>
+
+<b>注：以上精度指标的评估集是自建的版面区域检测数据集，包含中英文论文、杂志和研报等常见的 500 张文档类型图片。</b>
+
 * <b>表格版面检测模型</b>
 <table>
 <thead>

paddlex/configs/modules/layout_detection/PP-DocBlockLayout.yaml

@@ -0,0 +1,40 @@
+Global:
+  model: PP-DocBlockLayout
+  mode: check_dataset # check_dataset/train/evaluate/predict
+  dataset_dir: "/paddle/dataset/paddlex/layout/det_layout_examples"
+  device: gpu:0,1,2,3
+  output: "output"
+
+CheckDataset:
+  convert:
+    enable: False
+    src_dataset_type: null
+  split:
+    enable: False
+    train_percent: null
+    val_percent: null
+
+Train:
+  num_classes: 11
+  epochs_iters: 100
+  batch_size: 1
+  learning_rate: 0.0001
+  pretrain_weight_path: https://paddle-model-ecology.bj.bcebos.com/paddlex/official_pretrained_model/PP-DocBlockLayout_pretrain.pdparams
+  warmup_steps: 100
+  resume_path: null
+  log_interval: 10
+  eval_interval: 1
+
+Evaluate:
+  weight_path: "output/best_model/best_model.pdparams"
+  log_interval: 10
+
+Export:
+  weight_path: https://paddle-model-ecology.bj.bcebos.com/paddlex/official_pretrained_model/PP-DocBlockLayout_pretrain.pdparams
+
+Predict:
+  batch_size: 1
+  model_dir: "output/best_model/inference"
+  input: "https://paddle-model-ecology.bj.bcebos.com/paddlex/imgs/demo_image/layout.jpg"
+  kernel_option:
+    run_mode: paddle

paddlex/configs/modules/layout_detection/PP-DocLayout_plus-L.yaml

@@ -0,0 +1,40 @@
+Global:
+  model: PP-DocLayout_plus-L
+  mode: check_dataset # check_dataset/train/evaluate/predict
+  dataset_dir: "/paddle/dataset/paddlex/layout/det_layout_examples"
+  device: gpu:0,1,2,3
+  output: "output"
+
+CheckDataset:
+  convert:
+    enable: False
+    src_dataset_type: null
+  split:
+    enable: False
+    train_percent: null
+    val_percent: null
+
+Train:
+  num_classes: 11
+  epochs_iters: 100
+  batch_size: 1
+  learning_rate: 0.0001
+  pretrain_weight_path: https://paddle-model-ecology.bj.bcebos.com/paddlex/official_pretrained_model/PP-DocLayout_plus-L_pretrain.pdparams
+  warmup_steps: 100
+  resume_path: null
+  log_interval: 10
+  eval_interval: 1
+
+Evaluate:
+  weight_path: "output/best_model/best_model.pdparams"
+  log_interval: 10
+
+Export:
+  weight_path: https://paddle-model-ecology.bj.bcebos.com/paddlex/official_pretrained_model/PP-DocLayout_plus-L_pretrain.pdparams
+
+Predict:
+  batch_size: 1
+  model_dir: "output/best_model/inference"
+  input: "https://paddle-model-ecology.bj.bcebos.com/paddlex/imgs/demo_image/layout.jpg"
+  kernel_option:
+    run_mode: paddle

paddlex/inference/models/object_detection/predictor.py

@@ -316,6 +316,8 @@ class DetPredictor(BasePredictor):
             "BlazeFace",
             "BlazeFace-FPN-SSH",
             "PP-DocLayout-L",
+            "PP-DocLayout_plus-L",
+            "PP-DocBlockLayout",
         ]
         if any(name in self.model_name for name in models_required_imgsize):
             ordered_required_keys = (

paddlex/inference/models/object_detection/utils.py

@@ -65,4 +65,6 @@ STATIC_SHAPE_MODEL_LIST = [
     "PP-DocLayout-L",
     "PP-DocLayout-M",
     "PP-DocLayout-S",
+    "PP-DocLayout_plus-L",
+    "PP-DocBlockLayout",
 ]

paddlex/inference/utils/hpi_model_info_collection.json

@@ -445,6 +445,14 @@
       "onnxruntime",
       "paddle"
     ],
+    "PP-DocLayout_plus-L": [
+      "onnxruntime",
+      "paddle"
+    ],
+    "PP-DocBlockLayout": [
+      "onnxruntime",
+      "paddle"
+    ],
     "RT-DETR-H_layout_17cls": [
       "onnxruntime",
       "paddle"

paddlex/inference/utils/official_models.py

@@ -331,6 +331,8 @@ PP-LCNet_x1_0_vehicle_attribute_infer.tar",
     "PP-DocLayout-L": "https://paddle-model-ecology.bj.bcebos.com/paddlex/official_inference_model/paddle3.0.0/PP-DocLayout-L_infer.tar",
     "PP-DocLayout-M": "https://paddle-model-ecology.bj.bcebos.com/paddlex/official_inference_model/paddle3.0.0/PP-DocLayout-M_infer.tar",
     "PP-DocLayout-S": "https://paddle-model-ecology.bj.bcebos.com/paddlex/official_inference_model/paddle3.0.0/PP-DocLayout-S_infer.tar",
+    "PP-DocLayout_plus-L": "https://paddle-model-ecology.bj.bcebos.com/paddlex/official_inference_model/paddle3.0.0/PP-DocLayout_plus-L_infer.tar",
+    "PP-DocBlockLayout": "https://paddle-model-ecology.bj.bcebos.com/paddlex/official_inference_model/paddle3.0.0/PP-DocBlockLayout_infer.tar",
     "BEVFusion": "https://paddle-model-ecology.bj.bcebos.com/paddlex/official_inference_model/paddle3.0.0/BEVFusion_infer.tar",
     "YOLO-Worldv2-L": "https://paddle-model-ecology.bj.bcebos.com/paddlex/official_inference_model/paddle3.0.0/YOLO-Worldv2-L_infer.tar",
     "PP-DocBee-2B": "https://paddle-model-ecology.bj.bcebos.com/paddlex/official_inference_model/paddle3.0.0/PP-DocBee-2B_infer.tar",

paddlex/modules/object_detection/model_list.py

@@ -81,4 +81,6 @@ MODELS = [
     "PP-DocLayout-L",
     "PP-DocLayout-M",
     "PP-DocLayout-S",
+    "PP-DocLayout_plus-L",
+    "PP-DocBlockLayout",
 ]

paddlex/repo_apis/PaddleDetection_api/configs/PP-DocBlockLayout.yaml

@@ -0,0 +1,173 @@
+# Runtime
+epoch: 40
+log_iter: 10
+find_unused_parameters: true
+use_gpu: true
+use_xpu: false
+use_mlu: false
+use_npu: false
+use_ema: true
+ema_decay: 0.9999
+ema_decay_type: "exponential"
+ema_filter_no_grad: true
+save_dir: output
+snapshot_epoch: 1
+print_flops: false
+print_params: false
+eval_size: [640, 640]
+
+# Dataset
+metric: COCO
+num_classes: 1
+
+worker_num: 4
+
+TrainDataset:
+  name: COCODetDataset
+  image_dir: images
+  anno_path: annotations/instance_train.json
+  dataset_dir: datasets/COCO
+  data_fields: ['image', 'gt_bbox', 'gt_class', 'is_crowd']
+
+EvalDataset:
+  name: COCODetDataset
+  image_dir: images
+  anno_path: annotations/instance_val.json
+  dataset_dir: datasets/COCO
+  allow_empty: true
+
+TestDataset:
+  name: ImageFolder
+  anno_path: annotations/instance_val.json
+  dataset_dir: datasets/COCO
+
+TrainReader:
+  sample_transforms:
+    - Decode: {}
+    - RandomDistort: {prob: 0.8}
+    - RandomExpand: {fill_value: [123.675, 116.28, 103.53]}
+    - RandomCrop: {prob: 0.8}
+    - RandomFlip: {}
+  batch_transforms:
+    - BatchRandomResize: {target_size: [480, 512, 544, 576, 608, 640, 640, 640, 672, 704, 736, 768, 800], random_size: True, random_interp: True, keep_ratio: False}
+    - NormalizeImage: {mean: [0., 0., 0.], std: [1., 1., 1.], norm_type: none}
+    - NormalizeBox: {}
+    - BboxXYXY2XYWH: {}
+    - Permute: {}
+  batch_size: 8
+  shuffle: true
+  drop_last: true
+  collate_batch: false
+  use_shared_memory: true
+
+EvalReader:
+  sample_transforms:
+    - Decode: {}
+    - Resize: {target_size: [640, 640], keep_ratio: False, interp: 2}
+    - NormalizeImage: {mean: [0., 0., 0.], std: [1., 1., 1.], norm_type: none}
+    - Permute: {}
+  batch_size: 4
+  shuffle: false
+  drop_last: false
+
+TestReader:
+  inputs_def:
+    image_shape: [3, 640, 640]
+  sample_transforms:
+    - Decode: {}
+    - Resize: {target_size: [640, 640], keep_ratio: False, interp: 2}
+    - NormalizeImage: {mean: [0., 0., 0.], std: [1., 1., 1.], norm_type: none}
+    - Permute: {}
+  batch_size: 1
+  shuffle: false
+  drop_last: false
+
+# Model
+architecture: DETR
+pretrain_weights: https://paddledet.bj.bcebos.com/models/pretrained/PPHGNetV2_L_ssld_pretrained.pdparams
+
+norm_type: sync_bn
+hidden_dim: 256
+use_focal_loss: True
+
+DETR:
+  backbone: PPHGNetV2
+  neck: HybridEncoder
+  transformer: RTDETRTransformer
+  detr_head: DINOHead
+  post_process: DETRPostProcess
+
+PPHGNetV2:
+  arch: 'L'
+  return_idx: [1, 2, 3]
+  freeze_stem_only: true
+  freeze_at: 0
+  freeze_norm: true
+  lr_mult_list: [0., 0.05, 0.05, 0.05, 0.05]
+
+HybridEncoder:
+  hidden_dim: 256
+  use_encoder_idx: [2]
+  num_encoder_layers: 1
+  encoder_layer:
+    name: TransformerLayer
+    d_model: 256
+    nhead: 8
+    dim_feedforward: 1024
+    dropout: 0.
+    activation: 'gelu'
+  expansion: 1.0
+
+RTDETRTransformer:
+  num_queries: 300
+  position_embed_type: sine
+  feat_strides: [8, 16, 32]
+  num_levels: 3
+  nhead: 8
+  num_decoder_layers: 6
+  dim_feedforward: 1024
+  dropout: 0.0
+  activation: relu
+  num_denoising: 100
+  label_noise_ratio: 0.5
+  box_noise_scale: 1.0
+  learnt_init_query: false
+
+DINOHead:
+  loss:
+    name: DINOLoss
+    loss_coeff: {class: 1, bbox: 5, giou: 2}
+    aux_loss: true
+    use_vfl: true
+    matcher:
+      name: HungarianMatcher
+      matcher_coeff: {class: 2, bbox: 5, giou: 2}
+
+DETRPostProcess:
+  num_top_queries: 300
+
+# Optimizer
+LearningRate:
+  base_lr: 0.0001
+  schedulers:
+  - !PiecewiseDecay
+    gamma: 1.0
+    milestones: [100]
+    use_warmup: true
+  - !LinearWarmup
+    start_factor: 0.001
+    steps: 100
+
+OptimizerBuilder:
+  clip_grad_by_norm: 0.1
+  regularizer: false
+  optimizer:
+    type: AdamW
+    weight_decay: 0.0001
+
+# Export
+export:
+  post_process: true
+  nms: true
+  benchmark: false
+  fuse_conv_bn: false

paddlex/repo_apis/PaddleDetection_api/configs/PP-DocLayout_plus-L.yaml

@@ -0,0 +1,173 @@
+# Runtime
+epoch: 40
+log_iter: 10
+find_unused_parameters: true
+use_gpu: true
+use_xpu: false
+use_mlu: false
+use_npu: false
+use_ema: true
+ema_decay: 0.9999
+ema_decay_type: "exponential"
+ema_filter_no_grad: true
+save_dir: output
+snapshot_epoch: 1
+print_flops: false
+print_params: false
+eval_size: [800, 800]
+
+# Dataset
+metric: COCO
+num_classes: 23
+
+worker_num: 4
+
+TrainDataset:
+  name: COCODetDataset
+  image_dir: images
+  anno_path: annotations/instance_train.json
+  dataset_dir: datasets/COCO
+  data_fields: ['image', 'gt_bbox', 'gt_class', 'is_crowd']
+
+EvalDataset:
+  name: COCODetDataset
+  image_dir: images
+  anno_path: annotations/instance_val.json
+  dataset_dir: datasets/COCO
+  allow_empty: true
+
+TestDataset:
+  name: ImageFolder
+  anno_path: annotations/instance_val.json
+  dataset_dir: datasets/COCO
+
+TrainReader:
+  sample_transforms:
+    - Decode: {}
+    - RandomDistort: {prob: 0.8}
+    - RandomExpand: {fill_value: [123.675, 116.28, 103.53]}
+    - RandomCrop: {prob: 0.8}
+    - RandomFlip: {}
+  batch_transforms:
+    - BatchRandomResize: {target_size: [672, 704, 736, 768, 800, 800, 800, 800, 832, 864, 896, 928], random_size: True, random_interp: True, keep_ratio: False}
+    - NormalizeImage: {mean: [0., 0., 0.], std: [1., 1., 1.], norm_type: none}
+    - NormalizeBox: {}
+    - BboxXYXY2XYWH: {}
+    - Permute: {}
+  batch_size: 8
+  shuffle: true
+  drop_last: true
+  collate_batch: false
+  use_shared_memory: true
+
+EvalReader:
+  sample_transforms:
+    - Decode: {}
+    - Resize: {target_size: [800, 800], keep_ratio: False, interp: 2}
+    - NormalizeImage: {mean: [0., 0., 0.], std: [1., 1., 1.], norm_type: none}
+    - Permute: {}
+  batch_size: 4
+  shuffle: false
+  drop_last: false
+
+TestReader:
+  inputs_def:
+    image_shape: [3, 800, 800]
+  sample_transforms:
+    - Decode: {}
+    - Resize: {target_size: [800, 800], keep_ratio: False, interp: 2}
+    - NormalizeImage: {mean: [0., 0., 0.], std: [1., 1., 1.], norm_type: none}
+    - Permute: {}
+  batch_size: 1
+  shuffle: false
+  drop_last: false
+
+# Model
+architecture: DETR
+pretrain_weights: https://paddledet.bj.bcebos.com/models/pretrained/PPHGNetV2_L_ssld_pretrained.pdparams
+
+norm_type: sync_bn
+hidden_dim: 256
+use_focal_loss: True
+
+DETR:
+  backbone: PPHGNetV2
+  neck: HybridEncoder
+  transformer: RTDETRTransformer
+  detr_head: DINOHead
+  post_process: DETRPostProcess
+
+PPHGNetV2:
+  arch: 'L'
+  return_idx: [1, 2, 3]
+  freeze_stem_only: true
+  freeze_at: 0
+  freeze_norm: true
+  lr_mult_list: [0., 0.05, 0.05, 0.05, 0.05]
+
+HybridEncoder:
+  hidden_dim: 256
+  use_encoder_idx: [2]
+  num_encoder_layers: 1
+  encoder_layer:
+    name: TransformerLayer
+    d_model: 256
+    nhead: 8
+    dim_feedforward: 1024
+    dropout: 0.
+    activation: 'gelu'
+  expansion: 1.0
+
+RTDETRTransformer:
+  num_queries: 300
+  position_embed_type: sine
+  feat_strides: [8, 16, 32]
+  num_levels: 3
+  nhead: 8
+  num_decoder_layers: 6
+  dim_feedforward: 1024
+  dropout: 0.0
+  activation: relu
+  num_denoising: 100
+  label_noise_ratio: 0.5
+  box_noise_scale: 1.0
+  learnt_init_query: false
+
+DINOHead:
+  loss:
+    name: DINOLoss
+    loss_coeff: {class: 1, bbox: 5, giou: 2}
+    aux_loss: true
+    use_vfl: true
+    matcher:
+      name: HungarianMatcher
+      matcher_coeff: {class: 2, bbox: 5, giou: 2}
+
+DETRPostProcess:
+  num_top_queries: 300
+
+# Optimizer
+LearningRate:
+  base_lr: 0.0001
+  schedulers:
+  - !PiecewiseDecay
+    gamma: 1.0
+    milestones: [100]
+    use_warmup: true
+  - !LinearWarmup
+    start_factor: 0.001
+    steps: 100
+
+OptimizerBuilder:
+  clip_grad_by_norm: 0.1
+  regularizer: false
+  optimizer:
+    type: AdamW
+    weight_decay: 0.0001
+
+# Export
+export:
+  post_process: true
+  nms: true
+  benchmark: false
+  fuse_conv_bn: false

paddlex/repo_apis/PaddleDetection_api/object_det/official_categories.py

@@ -217,4 +217,29 @@ official_categories = {
         {"name": "footer_image", "id": 21},
         {"name": "aside_text", "id": 22},
     ],
+    "PP-DocLayout_plus-L": [
+        {"name": "paragraph_title", "id": 0},
+        {"name": "image", "id": 1},
+        {"name": "text", "id": 2},
+        {"name": "number", "id": 3},
+        {"name": "abstract", "id": 4},
+        {"name": "content", "id": 5},
+        {"name": "figure_title", "id": 6},
+        {"name": "formula", "id": 7},
+        {"name": "table", "id": 8},
+        {"name": "reference", "id": 9},
+        {"name": "doc_title", "id": 10},
+        {"name": "footnote", "id": 11},
+        {"name": "header", "id": 12},
+        {"name": "algorithm", "id": 13},
+        {"name": "footer", "id": 14},
+        {"name": "seal", "id": 15},
+        {"name": "chart", "id": 16},
+        {"name": "formula_number", "id": 17},
+        {"name": "aside_text", "id": 18},
+        {"name": "reference_content", "id": 19},
+    ],
+    "PP-DocBlockLayout": [
+        {"name": "Region", "id": 0},
+    ],
 }

paddlex/repo_apis/PaddleDetection_api/object_det/register.py

@@ -1103,3 +1103,33 @@ register_model_info(
         },
     }
 )
+
+register_model_info(
+    {
+        "model_name": "PP-DocLayout_plus-L",
+        "suite": "Det",
+        "config_path": osp.join(PDX_CONFIG_DIR, "PP-DocLayout_plus-L.yaml"),
+        "supported_apis": ["train", "evaluate", "predict", "export", "infer"],
+        "supported_dataset_types": ["COCODetDataset"],
+        "supported_train_opts": {
+            "device": ["cpu", "gpu_nxcx", "xpu", "npu", "mlu"],
+            "dy2st": False,
+            "amp": ["OFF"],
+        },
+    }
+)
+
+register_model_info(
+    {
+        "model_name": "PP-DocBlockLayout",
+        "suite": "Det",
+        "config_path": osp.join(PDX_CONFIG_DIR, "PP-DocBlockLayout.yaml"),
+        "supported_apis": ["train", "evaluate", "predict", "export", "infer"],
+        "supported_dataset_types": ["COCODetDataset"],
+        "supported_train_opts": {
+            "device": ["cpu", "gpu_nxcx", "xpu", "npu", "mlu"],
+            "dy2st": False,
+            "amp": ["OFF"],
+        },
+    }
+)