[MOT] update install dependency (#3726)

configs/ppyolo/README.md

@@ -60,6 +60,7 @@ PP-YOLO improved performance and speed of YOLOv3 with following methods:
 - PP-YOLO inference speed is tesed on single Tesla V100 with batch size as 1, CUDA 10.2, CUDNN 7.5.1, TensorRT 5.1.2.2 in TensorRT mode.
 - PP-YOLO FP32 inference speed testing uses inference model exported by `tools/export_model.py` and benchmarked by running `depoly/python/infer.py` with `--run_benchmark`. All testing results do not contains the time cost of data reading and post-processing(NMS), which is same as [YOLOv4(AlexyAB)](https://github.com/AlexeyAB/darknet) in testing method.
 - TensorRT FP16 inference speed testing exclude the time cost of bounding-box decoding(`yolo_box`) part comparing with FP32 testing above, which means that data reading, bounding-box decoding and post-processing(NMS) is excluded(test method same as [YOLOv4(AlexyAB)](https://github.com/AlexeyAB/darknet) too)
+- If you set `--run_benchmark=True`，you should install these dependencies at first, `pip install pynvml psutil GPUtil`.
 
 ### PP-YOLO for mobile
 

deploy/TENSOR_RT.md

@@ -31,6 +31,8 @@ config->EnableTensorRtEngine(1 << 20             /*workspace_size*/,
                              false             /*use_calib_mode*/);
 
 ```
+**注意:**
+  --run_benchmark如果设置为True，则需要安装依赖`pip install pynvml psutil GPUtil`。
 
 ### 3.2 TensorRT固定尺寸预测
 

deploy/cpp/docs/Jetson_build.md

@@ -168,7 +168,9 @@ CUDNN_LIB=/usr/lib/aarch64-linux-gnu/
 | --use_mkldnn | CPU预测中是否开启MKLDNN加速 |
 | --cpu_threads | 设置cpu线程数，默认为1 |
 
-**注意**: 优先级顺序：`camera_id` > `video_file` > `image_dir` > `image_file`。
+**注意**:
+- 优先级顺序：`camera_id` > `video_file` > `image_dir` > `image_file`。
+- --run_benchmark如果设置为True，则需要安装依赖`pip install pynvml psutil GPUtil`。
 
 
 `样例一`：

deploy/cpp/docs/linux_build.md

@@ -110,8 +110,9 @@ make
 | --use_mkldnn | CPU预测中是否开启MKLDNN加速 |
 | --cpu_threads | 设置cpu线程数，默认为1 |
 
-**注意**: 优先级顺序：`camera_id` > `video_file` > `image_dir` > `image_file`。
-
+**注意**:
+- 优先级顺序：`camera_id` > `video_file` > `image_dir` > `image_file`。
+- --run_benchmark如果设置为True，则需要安装依赖`pip install pynvml psutil GPUtil`。
 
 `样例一`：
 ```shell

deploy/cpp/docs/windows_vs2019_build.md

@@ -108,6 +108,7 @@ cd D:\projects\PaddleDetection\deploy\cpp\out\build\x64-Release
 **注意**：  
 （1）优先级顺序：`camera_id` > `video_file` > `image_dir` > `image_file`。
 （2）如果提示找不到`opencv_world346.dll`，把`D:\projects\packages\opencv3_4_6\build\x64\vc14\bin`文件夹下的`opencv_world346.dll`拷贝到`main.exe`文件夹下即可。
+（3）--run_benchmark如果设置为True，则需要安装依赖`pip install pynvml psutil GPUtil`。
 
 
 `样例一`：

deploy/python/README.md

@@ -48,3 +48,4 @@ python deploy/python/infer.py --model_dir=./inference/yolov3_mobilenet_v1_roadsi
 - 参数优先级顺序：`camera_id` > `video_file` > `image_dir` > `image_file`。
 - run_mode：fluid代表使用AnalysisPredictor，精度float32来推理，其他参数指用AnalysisPredictor，TensorRT不同精度来推理。
 - 如果安装的PaddlePaddle不支持基于TensorRT进行预测，需要自行编译，详细可参考[预测库编译教程](https://paddleinference.paddlepaddle.org.cn/user_guides/source_compile.html)。
+- --run_benchmark如果设置为True，则需要安装依赖`pip install pynvml psutil GPUtil`。

requirements.txt

@@ -14,3 +14,4 @@ lap
 sklearn
 motmetrics
 openpyxl
+cython_bbox