Add PP-Doclayout-V3

docs/model-zoo/ocr.md

@@ -18,3 +18,4 @@ hide:
 | [DocLayout-YOLO](https://github.com/opendatalab/DocLayout-YOLO) | Object Detection | [demo](https://github.com/jamjamjon/usls/tree/main/examples/ocr) | ✅ | ❓ | ✅ | ✅ | ✅ | ❌ | ❌ |
 | [PP-DocLayout-v1-Plus-L](https://huggingface.co/PaddlePaddle/PP-DocLayout_plus-L) | Object Detection | [demo](https://github.com/jamjamjon/usls/tree/main/examples/ocr) | ✅ | ✅ | ✅ | ✅ | ✅ | ✅ | ✅ |
 | [PP-DocLayout-v2](https://huggingface.co/PaddlePaddle/PP-DocLayoutV2) | Object Detection | [demo](https://github.com/jamjamjon/usls/tree/main/examples/ocr) | ✅ | ✅ | ✅ | ✅ | ✅ | ✅ | ✅ |
+| [PP-DocLayout-v3](https://huggingface.co/PaddlePaddle/PP-DocLayoutV3) | Object Detection | [demo](https://github.com/jamjamjon/usls/tree/main/examples/ocr) | ✅ | ✅ | ✅ | ✅ | ✅ | ✅ | ✅ |

examples/ocr/README.md

@@ -36,6 +36,10 @@ cargo run -F cuda-full -F vlm --example ocr -- picodet-layout --device cuda:0 --
 
 ## PP-DocLayout v1/v2
 cargo run -F cuda-full -F vlm --example ocr -- pp-doclayout --device cuda:0 --processor-device cuda:0 --source images/academic.jpg --ver 1 --dtype fp32
+
+
+## PP-DocLayout v3
+cargo run -F cuda-full -F vlm --example ocr -- pp-doclayout --device cuda:0 --processor-device cuda:0 --source images/vl1.58.png --ver 3 --dtype fp32
 ```
 
 ### Table structure recognition
@@ -53,3 +57,4 @@ cargo run -F cuda-full -F vlm --example ocr -- slanet --device cuda:0 --processo
 ![](https://github.com/jamjamjon/assets/releases/download/db/demo-table-ch.png)
 ![](https://github.com/jamjamjon/assets/releases/download/db/demo-sign.png)
 ![](https://github.com/jamjamjon/assets/releases/download/yolo/demo-doclayout-yolo.png)
+![](https://github.com/jamjamjon/assets/releases/download/pp-doclayout/demo-v3.jpg)

examples/ocr/main.rs

@@ -214,7 +214,21 @@ fn main() -> Result<()> {
         }
         Commands::PpDoclayout(args) => {
             let config = pp_doclayout::config(args)?.commit()?;
-            let annotator = Annotator::default();
+            let annotator = match args.ver.0 {
+                3 => Annotator::default()
+                    .with_hbb_style(
+                        usls::HbbStyle::default()
+                            .with_visible(false)
+                            .with_text_visible(false),
+                    )
+                    .with_mask_style(
+                        usls::MaskStyle::default()
+                            .with_visible(false)
+                            .with_cutout(true)
+                            .with_draw_obbs(true),
+                    ),
+                _ => Annotator::default(),
+            };
             run::<PPDocLayout>(config, &cli.source, &annotator, RUN_DOC_LAYOUT)
         }
     }?;

examples/ocr/pp_doclayout.rs

@@ -41,6 +41,7 @@ pub fn config(args: &PPDoclayoutArgs) -> Result<Config> {
     let config = match args.ver {
         Version(1, _, _) => Config::pp_doclayout_v1_plus_l(),
         Version(2, _, _) => Config::pp_doclayout_v2(),
+        Version(3, _, _) => Config::pp_doclayout_v3(),
         _ => anyhow::bail!("Unsupported version"),
     }
     .with_dtype_all(args.dtype)

src/models/vision/pp_doclayout/config.rs

@@ -6,6 +6,7 @@
 /// >
 /// > # Paper & Code
 /// >
+/// > - **v3 Model**: https://huggingface.co/PaddlePaddle/PP-DocLayoutV3
 /// > - **v2 Model**: https://huggingface.co/PaddlePaddle/PP-DocLayoutV2
 /// > - **v1 Model**: https://huggingface.co/PaddlePaddle/PP-DocLayout_plus-L
 /// > - **GitHub**: [PaddlePaddle/PaddleOCR](https://github.com/PaddlePaddle/PaddleOCR)
@@ -28,7 +29,7 @@ impl crate::Config {
             .with_model_ixx(1, 2, 800)
             .with_model_ixx(1, 3, 800)
             .with_model_ixx(2, 1, 2) // scale factors
-            .with_class_confs(&[0.5])
+            .with_class_confs(&[0.35])
             .with_resize_alg(crate::ResizeAlg::Interpolation(
                 crate::ResizeFilter::Bilinear,
             ))
@@ -50,4 +51,14 @@ impl crate::Config {
             .with_class_names(&crate::NAMES_PP_DOC_LAYOUT_V2_25)
             .with_model_file("v2.onnx")
     }
+
+    /// PP-DocLayoutV3 configuration (25 classes with reading order)
+    /// PP-DocLayoutV3 is specifically engineered to handle non-planar document images.
+    /// It can directly predict multi-point bounding boxes for layout elements—as opposed to standard two-point boxes—and determine logical reading orders for skewed and curved surfaces within a single forward pass, significantly reducing cascading errors.
+    pub fn pp_doclayout_v3() -> Self {
+        Self::pp_doclayout()
+            .with_version(3.into())
+            .with_class_names(&crate::NAMES_PP_DOC_LAYOUT_V2_25)
+            .with_model_file("v3.onnx")
+    }
 }

src/models/vision/pp_doclayout/impl.rs

@@ -5,10 +5,10 @@ use rayon::prelude::*;
 
 use crate::{
     elapsed_module, inputs, Config, DynConf, Engine, Engines, FromConfig, Hbb, Image,
-    ImageProcessor, Model, Module, Version, Xs, X, Y,
+    ImageProcessor, Mask, Model, Module, Ops, Version, Xs, X, Y,
 };
 
-/// RT-DETR: Real-Time Detection Transformer
+/// PP-DocLayoutsss-v1/v2/v3
 #[derive(Debug, Builder)]
 pub struct PPDocLayout {
     pub height: usize,
@@ -158,6 +158,68 @@ impl PPDocLayout {
 
                 Ok(ys)
             }
+            Version(3, _, _) => {
+                // preds: [batch * max_det, 7]，[label_index, score, xmin, ymin, xmax, ymax, reading_order]
+                let preds_reshaped = preds.to_shape([self.batch, self.max_det, 7])?;
+                let preds_masks = xs
+                    .get::<f32>(2)
+                    .ok_or_else(|| anyhow::anyhow!("Failed to get masks"))?;
+                let preds_masks = preds_masks.to_shape([self.batch, self.max_det, 200, 200])?;
+
+                let ys: Vec<Y> = preds_reshaped
+                    .axis_iter(Axis(0))
+                    .into_par_iter()
+                    .zip(preds_masks.axis_iter(Axis(0)).into_par_iter())
+                    .enumerate()
+                    .filter_map(|(idx, (preds, preds_masks))| {
+                        let info = &self.processor.images_transform_info[idx];
+                        let (image_height, image_width) = (info.height_src, info.width_src);
+
+                        let mut items: Vec<(Hbb, Mask, usize)> = preds
+                            .outer_iter()
+                            .zip(preds_masks.outer_iter())
+                            .filter_map(|(pred_slice, pred_mask)| {
+                                let slice = pred_slice.as_slice()?;
+                                let hbb = f(slice)?;
+                                let order = slice[6] as usize;
+                                let (mh, mw) = (pred_mask.shape()[0], pred_mask.shape()[1]);
+                                let (ih, iw) = (image_height, image_width);
+                                let mask_f32: Vec<f32> =
+                                    pred_mask.iter().map(|&x| 1. / (1. + (-x).exp())).collect();
+                                let mask_f32: Vec<f32> = Ops::interpolate_1d(
+                                    &mask_f32, mw as _, mh as _, iw as _, ih as _, false,
+                                )
+                                .ok()?;
+                                let mask_u8: Vec<u8> = mask_f32
+                                    .into_iter()
+                                    .map(|x| if x <= 0.5 { 0 } else { 1 })
+                                    .collect();
+
+                                let mut mask = Mask::new(&mask_u8, iw as _, ih as _).ok()?;
+
+                                if let Some(id) = hbb.id() {
+                                    mask = mask.with_id(id);
+                                }
+                                if let Some(name) = hbb.name() {
+                                    mask = mask.with_name(name);
+                                }
+                                if let Some(confidence) = hbb.confidence() {
+                                    mask = mask.with_confidence(confidence);
+                                }
+
+                                Some((hbb, mask, order))
+                            })
+                            .collect();
+
+                        items.sort_by(|a, b| a.2.cmp(&b.2));
+                        let hbbs: Vec<Hbb> = items.iter().map(|(h, _, _)| h.clone()).collect();
+                        let masks: Vec<Mask> = items.into_iter().map(|(_, m, _)| m).collect();
+                        Some(Y::default().with_hbbs(&hbbs).with_masks(&masks))
+                    })
+                    .collect();
+
+                Ok(ys)
+            }
             _ => anyhow::bail!("Unsupported version"),
         }
     }