update README

README.md

@@ -3,7 +3,7 @@
 
 该仓库基于 [shouxieai/tensorRT_Pro](https://github.com/shouxieai/tensorRT_Pro)，并进行了调整以支持 YOLOv8 的各项任务。
 
-* 目前已支持 YOLOv8、YOLOv8-Cls、YOLOv8-Seg、YOLOv8-OBB、YOLOv8-Pose、RT-DETR、ByteTrack 高性能推理！！！🚀🚀🚀
+* 目前已支持 YOLOv8、YOLOv8-Cls、YOLOv8-Seg、YOLOv8-OBB、YOLOv8-Pose、RT-DETR、ByteTrack、YOLOv9 高性能推理！！！🚀🚀🚀
 * 基于 tensorRT8.x，C++ 高级接口，C++ 部署，服务器/嵌入式使用
 
 <div align=center><img src="./assets/output.jpg" width="50%" height="50%"></div>
@@ -15,9 +15,12 @@
 - 🔥 [YOLOv8-OBB推理详解及部署实现](https://blog.csdn.net/qq_40672115/article/details/135713830)
 - 🔥 [YOLOv8-Pose推理详解及部署实现](https://blog.csdn.net/qq_40672115/article/details/134278117)
 - 🔥 [RT-DETR推理详解及部署实现](https://blog.csdn.net/qq_40672115/article/details/134356250)
+- 🔥 [YOLOv9推理详解及部署实现](https://blog.csdn.net/qq_40672115/article/details/136492338)
 
 
 ## Top News
+- **2024/3/5**
+  - YOLOv9 支持
 - **2024/2/1**
   - 新增 MinMaxCalibrator 校准器，可以通过 `TRT::Calibrator::MinMax` 指定
   - 新增 mAP 测试使用的一些脚本文件，mAP 计算代码 copy 自 [yolov6/core/evaler.py#L231](https://github.com/meituan/YOLOv6/blob/main/yolov6/core/evaler.py#L231)
@@ -793,6 +796,80 @@ make bytetrack -j64
 
 </details>
 
+<details>
+<summary>YOLOv9支持</summary>
+
+0. 说明
+
+本项目的 YOLOv9 部署实现并不是官方原版，而是采用的集成到 ultralytics 的 YOLOv9
+
+1. 下载 YOLOv8
+
+```shell
+git clone https://github.com/ultralytics/ultralytics.git
+```
+
+2. 修改代码, 保证动态 batch
+
+```python
+# ========== head.py ==========
+
+# ultralytics/nn/modules/head.py第75行，forward函数
+# return y if self.export else (y, x)
+# 修改为：
+
+return y.permute(0, 2, 1) if self.export else (y, x)
+
+# ========== exporter.py ==========
+
+# ultralytics/engine/exporter.py第365行
+# output_names = ['output0', 'output1'] if isinstance(self.model, SegmentationModel) else ['output0']
+# dynamic = self.args.dynamic
+# if dynamic:
+#     dynamic = {'images': {0: 'batch', 2: 'height', 3: 'width'}}  # shape(1,3,640,640)
+#     if isinstance(self.model, SegmentationModel):
+#         dynamic['output0'] = {0: 'batch', 2: 'anchors'}  # shape(1, 116, 8400)
+#         dynamic['output1'] = {0: 'batch', 2: 'mask_height', 3: 'mask_width'}  # shape(1,32,160,160)
+#     elif isinstance(self.model, DetectionModel):
+#         dynamic['output0'] = {0: 'batch', 2: 'anchors'}  # shape(1, 84, 8400)
+# 修改为：
+
+output_names = ['output0', 'output1'] if isinstance(self.model, SegmentationModel) else ['output']
+dynamic = self.args.dynamic
+if dynamic:
+    dynamic = {'images': {0: 'batch'}}  # shape(1,3,640,640)
+    if isinstance(self.model, SegmentationModel):
+        dynamic['output0'] = {0: 'batch', 2: 'anchors'}  # shape(1, 116, 8400)
+        dynamic['output1'] = {0: 'batch', 2: 'mask_height', 3: 'mask_width'}  # shape(1,32,160,160)
+    elif isinstance(self.model, DetectionModel):
+        dynamic['output'] = {0: 'batch'}  # shape(1, 84, 8400)
+```
+
+3. 导出 onnx 模型, 在 ultralytics-main 新建导出文件 `export.py` 内容如下：
+
+```python
+# ========== export.py ==========
+from ultralytics import YOLO
+
+model = YOLO("yolov9c.pt")
+
+success = model.export(format="onnx", dynamic=True, simplify=True)
+```
+
+```shell
+cd ultralytics-main
+python export.py
+```
+
+4. 复制模型并执行
+
+```shell
+cp ultralytics/yolov9c.onnx tensorRT_Pro-YOLOv8/workspace
+cd tensorRT_Pro-YOLOv8
+make yolo -j64
+```
+</details>
+
 ## 接口介绍
 
 <details>
@@ -840,4 +917,5 @@ auto boxes = engine->commit(image).get()  // 得到的是 vector<Box>
 
 ## 参考
 - [https://github.com/shouxieai/tensorRT_Pro](https://github.com/shouxieai/tensorRT_Pro)
-- [https://github.com/shouxieai/infer](https://github.com/shouxieai/infer)
+- [https://github.com/shouxieai/infer](https://github.com/shouxieai/infer)
+- [https://github.com/ultralytics/ultralytics](https://github.com/ultralytics/ultralytics)