update ppvehicle modelzoo (#6679)

* update ppvehicle modelzoo; test=document_fix * update pipeline pictures * update * fix platerec jetson trtfp16; test=document_fix * replace all --model_dir in pipeline docs; test=document_fix

update ppvehicle modelzoo (#6679)
* update ppvehicle modelzoo; test=document_fix * update pipeline pictures * update * fix platerec jetson trtfp16; test=document_fix * replace all --model_dir in pipeline docs; test=document_fix
6eff0446 · zhiboniu · GitHub · 8fbdf1cb · 6eff0446 · 6eff0446
15 changed file
--- a/deploy/pipeline/README.md
+++ b/deploy/pipeline/README.md
@@ -71,20 +71,14 @@
 <details>
 <summary><b>端到端模型效果（点击展开）</b></summary>
-| 任务        | 端到端速度（ms） | 模型方案                                                                                                                                                                                                                                                                  | 模型体积                                        |
+| 任务            | 端到端速度（ms）|  模型方案  |  模型体积 |
-| --------- | --------- | --------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | ------------------------------------------- |
+| :---------:     | :-------:  |  :------: |:------: |
-| 行人检测（高精度） | 25.1ms    | [多目标跟踪](https://bj.bcebos.com/v1/paddledet/models/pipeline/mot_ppyoloe_l_36e_pipeline.zip)                                                                                                                                                                            | 182M                                        |
+|  车辆检测（高精度）  | 25.7ms  |  [多目标跟踪](https://bj.bcebos.com/v1/paddledet/models/pipeline/mot_ppyoloe_l_36e_ppvehicle.zip) | 182M |  
-| 行人检测（轻量级） | 16.2ms    | [多目标跟踪](https://bj.bcebos.com/v1/paddledet/models/pipeline/mot_ppyoloe_s_36e_pipeline.zip)                                                                                                                                                                            | 27M                                         |
+|  车辆检测（轻量级）  | 13.2ms  |  [多目标跟踪](https://bj.bcebos.com/v1/paddledet/models/pipeline/mot_ppyoloe_s_36e_ppvehicle.zip) | 27M |
-| 行人跟踪（高精度） | 31.8ms    | [多目标跟踪](https://bj.bcebos.com/v1/paddledet/models/pipeline/mot_ppyoloe_l_36e_pipeline.zip)                                                                                                                                                                            | 182M                                        |
+|  车辆跟踪（高精度）  | 40ms  |  [多目标跟踪](https://bj.bcebos.com/v1/paddledet/models/pipeline/mot_ppyoloe_l_36e_ppvehicle.zip) | 182M |
-| 行人跟踪（轻量级） | 21.0ms    | [多目标跟踪](https://bj.bcebos.com/v1/paddledet/models/pipeline/mot_ppyoloe_s_36e_pipeline.zip)                                                                                                                                                                            | 27M                                         |
+|  车辆跟踪（轻量级）  | 25ms  |  [多目标跟踪](https://bj.bcebos.com/v1/paddledet/models/pipeline/mot_ppyoloe_s_36e_ppvehicle.zip) | 27M |
-| 属性识别（高精度） | 单人8.5ms   | [目标检测](https://bj.bcebos.com/v1/paddledet/models/pipeline/mot_ppyoloe_l_36e_pipeline.zip)<br> [属性识别](https://bj.bcebos.com/v1/paddledet/models/pipeline/strongbaseline_r50_30e_pa100k.zip)                                                                            | 目标检测：182M<br>属性识别：86M                       |
+|  车牌识别  |   4.68ms |  [车牌检测](https://bj.bcebos.com/v1/paddledet/models/pipeline/ch_PP-OCRv3_det_infer.tar.gz) <br> [车牌字符识别](https://bj.bcebos.com/v1/paddledet/models/pipeline/ch_PP-OCRv3_rec_infer.tar.gz) | 车牌检测：3.9M  <br> 车牌字符识别： 12M |
-| 属性识别（轻量级） | 单人7.1ms   | [目标检测](https://bj.bcebos.com/v1/paddledet/models/pipeline/mot_ppyoloe_l_36e_pipeline.zip)<br> [属性识别](https://bj.bcebos.com/v1/paddledet/models/pipeline/strongbaseline_r50_30e_pa100k.zip)                                                                            | 目标检测：182M<br>属性识别：86M                       |
+|  车辆属性  |   7.31ms | [车辆属性](https://bj.bcebos.com/v1/paddledet/models/pipeline/vehicle_attribute_model.zip) | 7.2M |
-| 摔倒识别      | 单人10ms    | [多目标跟踪](https://bj.bcebos.com/v1/paddledet/models/pipeline/mot_ppyoloe_l_36e_pipeline.zip) <br> [关键点检测](https://bj.bcebos.com/v1/paddledet/models/pipeline/dark_hrnet_w32_256x192.zip) <br> [基于关键点行为识别](https://bj.bcebos.com/v1/paddledet/models/pipeline/STGCN.zip) | 多目标跟踪：182M<br>关键点检测：101M<br>基于关键点行为识别：21.8M |
-| 闯入识别      | 31.8ms    | [多目标跟踪](https://bj.bcebos.com/v1/paddledet/models/pipeline/mot_ppyoloe_l_36e_pipeline.zip)                                                                                                                                                                            | 182M                                        |
-| 打架识别      | 19.7ms    | [视频分类](https://bj.bcebos.com/v1/paddledet/models/pipeline/mot_ppyoloe_l_36e_pipeline.zip)                                                                                                                                                                             | 90M                                         |
-| 抽烟识别      | 单人15.1ms  | [目标检测](https://bj.bcebos.com/v1/paddledet/models/pipeline/mot_ppyoloe_l_36e_pipeline.zip)<br>[基于人体id的目标检测](https://bj.bcebos.com/v1/paddledet/models/pipeline/ppyoloe_crn_s_80e_smoking_visdrone.zip)                                                                 | 目标检测：182M<br>基于人体id的目标检测：27M                |
-| 打电话识别     | 单人ms      | [目标检测](https://bj.bcebos.com/v1/paddledet/models/pipeline/mot_ppyoloe_l_36e_pipeline.zip)<br>[基于人体id的图像分类](https://bj.bcebos.com/v1/paddledet/models/pipeline/PPHGNet_tiny_calling_halfbody.zip)                                                                      | 目标检测：182M<br>基于人体id的图像分类：45M                |
 点击模型方案中的模型即可下载指定模型，下载后解压存放至`./output_inference`目录中

--- a/deploy/pipeline/docs/tutorials/PPVehicle_QUICK_STARTED.md
+++ b/deploy/pipeline/docs/tutorials/PPVehicle_QUICK_STARTED.md
@@ -46,11 +46,13 @@ PP-Vehicle提供了目标检测、属性识别、行为识别、ReID预训练模
 | 任务            | 端到端速度（ms）|  模型方案  |  模型体积 |
 | :---------:     | :-------:  |  :------: |:------: |
-|  车辆检测（高精度）  | 25.1ms  |  [多目标跟踪](https://bj.bcebos.com/v1/paddledet/models/pipeline/mot_ppyoloe_l_36e_ppvehicle.zip) | 182M |  
+|  车辆检测（高精度）  | 25.7ms  |  [多目标跟踪](https://bj.bcebos.com/v1/paddledet/models/pipeline/mot_ppyoloe_l_36e_ppvehicle.zip) | 182M |  
-|  车辆检测（轻量级）  | 16.2ms  |  [多目标跟踪](https://bj.bcebos.com/v1/paddledet/models/pipeline/mot_ppyoloe_s_36e_ppvehicle.zip) | 27M |
+|  车辆检测（轻量级）  | 13.2ms  |  [多目标跟踪](https://bj.bcebos.com/v1/paddledet/models/pipeline/mot_ppyoloe_s_36e_ppvehicle.zip) | 27M |
-|  车辆跟踪（高精度）  | 31.8ms  |  [多目标跟踪](https://bj.bcebos.com/v1/paddledet/models/pipeline/mot_ppyoloe_l_36e_ppvehicle.zip) | 182M |
+|  车辆跟踪（高精度）  | 40ms  |  [多目标跟踪](https://bj.bcebos.com/v1/paddledet/models/pipeline/mot_ppyoloe_l_36e_ppvehicle.zip) | 182M |
-|  车辆跟踪（轻量级）  | 21.0ms  |  [多目标跟踪](https://bj.bcebos.com/v1/paddledet/models/pipeline/mot_ppyoloe_s_36e_ppvehicle.zip) | 27M |
+|  车辆跟踪（轻量级）  | 25ms  |  [多目标跟踪](https://bj.bcebos.com/v1/paddledet/models/pipeline/mot_ppyoloe_s_36e_ppvehicle.zip) | 27M |
-|  闯入识别  |   31.8ms | [多目标跟踪](https://bj.bcebos.com/v1/paddledet/models/pipeline/mot_ppyoloe_l_36e_ppvehicle.zip) | 多目标跟踪：182M |
+|  车牌识别  |   4.68ms |  [车牌检测](https://bj.bcebos.com/v1/paddledet/models/pipeline/ch_PP-OCRv3_det_infer.tar.gz) <br> [车牌字符识别](https://bj.bcebos.com/v1/paddledet/models/pipeline/ch_PP-OCRv3_rec_infer.tar.gz) | 车牌检测：3.9M  <br> 车牌字符识别： 12M |
+|  车辆属性  |   7.31ms | [车辆属性](https://bj.bcebos.com/v1/paddledet/models/pipeline/vehicle_attribute_model.zip) | 7.2M |
 下载模型后，解压至`./output_inference`文件夹。
@@ -97,7 +99,7 @@ VEHICLE_ATTR:
 **注意：**
 - 如果用户需要实现不同任务，可以在配置文件对应enable选项设置为True。
- 如果用户仅需要修改模型文件路径，可以在命令行中加入 `--model_dir det=ppyoloe/` 即可，也可以手动修改配置文件中的相应模型路径，详细说明参考下方参数说明文档。
+- 如果用户仅需要修改模型文件路径，可以在命令行中--config后面紧跟着 `-o MOT.model_dir=ppyoloe/` 进行修改即可，也可以手动修改配置文件中的相应模型路径，详细说明参考下方参数说明文档。
 ## 预测部署
@@ -111,7 +113,7 @@ python deploy/pipeline/pipeline.py --config deploy/pipeline/config/infer_cfg_ppv
 # 车辆跟踪，指定配置文件路径，模型路径和测试视频，在配置文件```deploy/pipeline/config/infer_cfg_ppvehicle.yml```中的MOT部分enable设置为```True```
 # 命令行中指定的模型路径优先级高于配置文件
-python deploy/pipeline/pipeline.py --config deploy/pipeline/config/infer_cfg_ppvehicle.yml --video_file=test_video.mp4 --device=gpu --model_dir det=ppyoloe/ [--run_mode trt_fp16]
+python deploy/pipeline/pipeline.py --config deploy/pipeline/config/infer_cfg_ppvehicle.yml --video_file=test_video.mp4 --device=gpu  [--run_mode trt_fp16]
 # 车辆属性识别，指定配置文件路径和测试视频，在配置文件```deploy/pipeline/config/infer_cfg_ppvehicle.yml```中的ATTR部分enable设置为```True```
 python deploy/pipeline/pipeline.py --config deploy/pipeline/config/infer_cfg_ppvehicle.yml --video_file=test_video.mp4 --device=gpu [--run_mode trt_fp16]
@@ -130,7 +132,6 @@ python deploy/pipeline/pipeline.py --config deploy/pipeline/config/infer_cfg_ppv
 |-------|-------|----------|
 | --config | Yes | 配置文件路径 |
 | -o | Option | 覆盖配置文件中对应的配置  |
-| --model_dir | Option | 各任务模型路径，优先级高于配置文件, 例如`--model_dir det=better_det/ attr=better_attr/`|
 | --image_file | Option | 需要预测的图片 |
 | --image_dir  | Option |  要预测的图片文件夹路径   |
 | --video_file | Option | 需要预测的视频，或者rtsp流地址 |

--- a/deploy/pipeline/docs/tutorials/pphuman_action.md
+++ b/deploy/pipeline/docs/tutorials/pphuman_action.md
@@ -60,12 +60,12 @@ python deploy/pipeline/pipeline.py --config deploy/pipeline/config/infer_cfg_pph
 3. 若修改模型路径，有以下两种方式：
    - ```./deploy/pipeline/config/infer_cfg_pphuman.yml```下可以配置不同模型路径，关键点模型和摔倒行为识别模型分别对应`KPT`和`SKELETON_ACTION`字段，修改对应字段下的路径为实际期望的路径即可。
-    - 命令行中增加`--model_dir`修改模型路径：
+    - 命令行中--config后面紧跟着增加`-o KPT.model_dir=xxx SKELETON_ACTION.model_dir=xxx `修改模型路径：
 ```python
 python deploy/pipeline/pipeline.py --config deploy/pipeline/config/infer_cfg_pphuman.yml \
+                                                   -o KPT.model_dir=./dark_hrnet_w32_256x192 SKELETON_ACTION.model_dir=./STGCN\
                                                   --video_file=test_video.mp4 \
-                                                   --device=gpu \
+                                                   --device=gpu
-                                                   --model_dir kpt=./dark_hrnet_w32_256x192 action=./STGCN
 ```
 4. 启动命令中的完整参数说明，请参考[参数说明](./PPHuman_QUICK_STARTED.md)。

--- a/deploy/pipeline/docs/tutorials/pphuman_action_en.md
+++ b/deploy/pipeline/docs/tutorials/pphuman_action_en.md
@@ -72,13 +72,13 @@ SKELETON_ACTION: # Config for skeleton-based action recognition model
 3. There are two ways to modify the model path:
  - In ```./deploy/pipeline/config/infer_cfg_pphuman.yml```, you can configurate different model paths，which is proper only if you match keypoint models and action recognition models with the fields of `KPT` and `SKELETON_ACTION` respectively, and modify the corresponding path of each field into the expected path.
-  - Add `--model_dir` in the command line to revise the model path：
+  - Add `-o KPT.model_dir=xxx SKELETON_ACTION.model_dir=xxx ` in the command line following the --config to change the model path：
    ```python
-    python deploy/pipeline/pipeline.py --config deploy/pipeline/config/infer_cfg_pphuman.yml \
+python deploy/pipeline/pipeline.py --config deploy/pipeline/config/infer_cfg_pphuman.yml \
+                                                   -o KPT.model_dir=./dark_hrnet_w32_256x192 SKELETON_ACTION.model_dir=./STGCN\
                                                   --video_file=test_video.mp4 \
-                                                       --device=gpu \
+                                                   --device=gpu
-                                                       --model_dir kpt=./dark_hrnet_w32_256x192 action=./STGCN
    ```
 4. For detailed parameter description, please refer to [Parameter Description](./PPHuman_QUICK_STARTED.md)

--- a/deploy/pipeline/docs/tutorials/pphuman_attribute.md
+++ b/deploy/pipeline/docs/tutorials/pphuman_attribute.md
@@ -59,12 +59,12 @@ python deploy/pipeline/pipeline.py --config deploy/pipeline/config/infer_cfg_pph
 4. 若修改模型路径，有以下两种方式：
    - 方法一：```./deploy/pipeline/config/infer_cfg_pphuman.yml```下可以配置不同模型路径，属性识别模型修改ATTR字段下配置
-    - 方法二：命令行中增加`--model_dir`修改模型路径：
+    - 方法二：命令行中--config后面紧跟着增加`-o ATTR.model_dir`修改模型路径：
 ```python
-python deploy/pipeline/pipeline.py --config deploy/pipeline/config/infer_cfg_pphuman.yml \
+python deploy/pipeline/pipeline.py --config deploy/pipeline/config/infer_cfg_pphuman.yml
+                                                   -o ATTR.model_dir=output_inference/PPLCNet_x1_0_person_attribute_945_infer/\
                                                   --video_file=test_video.mp4 \
-                                                   --device=gpu \
+                                                   --device=gpu
-                                                   --model_dir attr=output_inference/PPLCNet_x1_0_person_attribute_945_infer/
 ```
 测试效果如下：

--- a/deploy/pipeline/docs/tutorials/pphuman_attribute_en.md
+++ b/deploy/pipeline/docs/tutorials/pphuman_attribute_en.md
@@ -55,12 +55,12 @@ python deploy/pipeline/pipeline.py --config deploy/pipeline/config/infer_cfg_pph
 4. If you want to change the model path, there are two methods：
    - The first: In ```./deploy/pipeline/config/infer_cfg_pphuman.yml``` you can configurate different model paths. In attribute recognition models, you can modify the configuration in the field of ATTR.
-    - The second: Add `--model_dir` in the command line to change the model path：
+    - The second: Add `-o ATTR.model_dir` in the command line following the --config to change the model path：
 ```python
 python deploy/pipeline/pipeline.py --config deploy/pipeline/config/infer_cfg_pphuman.yml \
+                                                   -o ATTR.model_dir=output_inference/PPLCNet_x1_0_person_attribute_945_infer/\
                                                   --video_file=test_video.mp4 \
-                                                   --device=gpu \
+                                                   --device=gpu
-                                                   --model_dir attr=output_inference/PPLCNet_x1_0_person_attribute_945_infer/
 ```
 The test result is：

--- a/deploy/pipeline/docs/tutorials/pphuman_mot.md
+++ b/deploy/pipeline/docs/tutorials/pphuman_mot.md
@@ -39,15 +39,15 @@ python deploy/pipeline/pipeline.py --config deploy/pipeline/config/infer_cfg_pph
 4. 若修改模型路径，有以下两种方式：
    - ```./deploy/pipeline/config/infer_cfg_pphuman.yml```下可以配置不同模型路径，检测和跟踪模型分别对应`DET`和`MOT`字段，修改对应字段下的路径为实际期望的路径即可。
-    - 命令行中增加`--model_dir`修改模型路径：
+    - 命令行中--config后面紧跟着增加`-o MOT.model_dir`修改模型路径：
 ```python
 python deploy/pipeline/pipeline.py --config deploy/pipeline/config/infer_cfg_pphuman.yml \
+                                                   -o MOT.model_dir=ppyoloe/\
                                                   --video_file=test_video.mp4 \
                                                   --device=gpu \
                                                   --region_type=horizontal \
                                                   --do_entrance_counting \
-                                                   --draw_center_traj \
+                                                   --draw_center_traj
-                                                   --model_dir det=ppyoloe/
 ```
 **注意:**

--- a/deploy/pipeline/docs/tutorials/pphuman_mot_en.md
+++ b/deploy/pipeline/docs/tutorials/pphuman_mot_en.md
@@ -39,16 +39,16 @@ python deploy/pipeline/pipeline.py --config deploy/pipeline/config/infer_cfg_pph
 4. There are two ways to modify the model path:
     - In `./deploy/pipeline/config/infer_cfg_pphuman.yml`, you can configurate different model paths，which is proper only if you match keypoint models and action recognition models with the fields of `DET` and `MOT` respectively, and modify the corresponding path of each field into the expected path.
-    - Add `--model_dir` in the command line to revise the model path:
+    - Add `-o MOT.model_dir` in the command line following the --config to change the model path：
 ```python
 python deploy/pipeline/pipeline.py --config deploy/pipeline/config/infer_cfg_pphuman.yml \
+                                                   -o MOT.model_dir=ppyoloe/\
                                                   --video_file=test_video.mp4 \
                                                   --device=gpu \
                                                   --region_type=horizontal \
                                                   --do_entrance_counting \
-                                                   --draw_center_traj \
+                                                   --draw_center_traj
-                                                   --model_dir det=ppyoloe/
 ```
 **Note:**

--- a/deploy/pipeline/docs/tutorials/pphuman_mtmct.md
+++ b/deploy/pipeline/docs/tutorials/pphuman_mtmct.md
@@ -18,10 +18,9 @@ python3 deploy/pipeline/pipeline.py --config deploy/pipeline/config/infer_cfg_pp
 ```python
 python3 deploy/pipeline/pipeline.py
-        --config deploy/pipeline/config/infer_cfg_pphuman.yml
+        --config deploy/pipeline/config/infer_cfg_pphuman.yml -o REID.model_dir=reid_best/
        --video_dir=[your_video_file_directory]
        --device=gpu
-        --model_dir reid=reid_best/
 ```
 ## 方案说明

--- a/deploy/pipeline/docs/tutorials/pphuman_mtmct_en.md
+++ b/deploy/pipeline/docs/tutorials/pphuman_mtmct_en.md
@@ -18,10 +18,9 @@ python3 deploy/pipeline/pipeline.py --config deploy/pipeline/config/infer_cfg_pp
 ```python
 python3 deploy/pipeline/pipeline.py
-        --config deploy/pipeline/config/infer_cfg_pphuman.yml
+        --config deploy/pipeline/config/infer_cfg_pphuman.yml -o REID.model_dir=reid_best/
        --video_dir=[your_video_file_directory]
        --device=gpu
-        --model_dir reid=reid_best/
 ```
 ## Intorduction to the Solution

--- a/deploy/pipeline/pipe_utils.py
+++ b/deploy/pipeline/pipe_utils.py
@@ -88,7 +88,8 @@ class PipeTimer(Times):
        for k, v in self.module_time.items():
            v_time = round(v.value(), 4)
            if v_time > 0 and k in ['det', 'mot', 'video_action']:
-                print("{} time(ms): {}".format(k, v_time * 1000))
+                print("{} time(ms): {}; per frame average time(ms): {}".format(
+                    k, v_time * 1000, v_time * 1000 / self.img_num))
            elif v_time > 0:
                print("{} time(ms): {}; per trackid average time(ms): {}".
                      format(k, v_time * 1000, v_time * 1000 / self.track_num))

--- a/deploy/pipeline/pipeline.py
+++ b/deploy/pipeline/pipeline.py
@@ -465,6 +465,7 @@ class PipePredictor(object):
                                                    self.cfg['crop_thresh'])
            if i > self.warmup_frame:
                self.pipe_timer.module_time['det'].end()
+                self.pipe_timer.track_num += len(det_res['boxes'])
            self.pipeline_res.update(det_res, 'det')
            if self.with_human_attr:

--- a/deploy/pipeline/ppvehicle/vehicle_plateutils.py
+++ b/deploy/pipeline/ppvehicle/vehicle_plateutils.py
@@ -139,6 +139,7 @@ def create_predictor(args, cfg, mode):
                min_input_shape = {"x": [1, 3, imgH, 10]}
                max_input_shape = {"x": [batch_size, 3, imgH, 2304]}
                opt_input_shape = {"x": [batch_size, 3, imgH, 320]}
+                config.exp_disable_tensorrt_ops(["transpose2"])
            elif mode == "cls":
                min_input_shape = {"x": [1, 3, 48, 10]}
                max_input_shape = {"x": [batch_size, 3, 48, 1024]}

--- a/docs/images/pphumanv2.png
+++ b/docs/images/pphumanv2.png
--- a/docs/images/ppvehicle.png
+++ b/docs/images/ppvehicle.png