diff --git a/configs/mot/README.md b/configs/mot/README.md
index b6f7615922a8b309d26f77490188b01c4afb227e..5497195644961c435cf59f5edd6c650a9244fe24 100644
--- a/configs/mot/README.md
+++ b/configs/mot/README.md
@@ -186,9 +186,21 @@ Evaluating the track performance of FairMOT on val dataset in single GPU with fo
 CUDA_VISIBLE_DEVICES=0 python tools/eval_mot.py -c configs/mot/fairmot/fairmot_dla34_30e_1088x608.yml -o weights=https://paddledet.bj.bcebos.com/models/mot/fairmot_dla34_30e_1088x608.pdparams
 
 # use saved checkpoint in training
-CUDA_VISIBLE_DEVICES=0 python tools/eval_mot.py -c configs/mot/fairmot/fairmot_dla34_30e_1088x608.yml -o weights=output/fairmot_dla34_30e_1088x608/model_final
+CUDA_VISIBLE_DEVICES=0 python tools/eval_mot.py -c configs/mot/fairmot/fairmot_dla34_30e_1088x608.yml -o weights=output/fairmot_dla34_30e_1088x608/model_final.pdparams
 ```
 
+### 3. Inference
+
+Inference a vidoe on single GPU with following command:
+
+```bash
+# inference on video and save a video
+CUDA_VISIBLE_DEVICES=0 python tools/infer_mot.py -c configs/mot/fairmot/fairmot_dla34_30e_1088x608.yml -o weights=https://paddledet.bj.bcebos.com/models/mot/fairmot_dla34_30e_1088x608.pdparams --video_file={your video name}.mp4  --save_videos
+```
+**Notes:**
+ Please make sure that [ffmpeg](https://ffmpeg.org/ffmpeg.html) is installed first, on Linux(Ubuntu) platform you can directly install it by the following command:`apt-get update && apt-get install -y ffmpeg`.
+
+
 ## Citations
 ```
 @article{wang2019towards,
diff --git a/configs/mot/README_cn.md b/configs/mot/README_cn.md
index 15fac03a2e0f9930ca4da7658482e87033def8d2..53f8e4db42fd4c79a569cf9d1b87618b67db798d 100644
--- a/configs/mot/README_cn.md
+++ b/configs/mot/README_cn.md
@@ -184,9 +184,21 @@ FairMOT使用单张GPU通过如下命令一键式启动评估
 CUDA_VISIBLE_DEVICES=0 python tools/eval_mot.py -c configs/mot/fairmot/fairmot_dla34_30e_1088x608.yml -o weights=https://paddledet.bj.bcebos.com/models/mot/fairmot_dla34_30e_1088x608.pdparams
 
 # 使用训练保存的checkpoint
-CUDA_VISIBLE_DEVICES=0 python tools/eval_mot.py -c configs/mot/fairmot/fairmot_dla34_30e_1088x608.yml -o weights=output/fairmot_dla34_30e_1088x608/model_final
+CUDA_VISIBLE_DEVICES=0 python tools/eval_mot.py -c configs/mot/fairmot/fairmot_dla34_30e_1088x608.yml -o weights=output/fairmot_dla34_30e_1088x608/model_final.pdparams
 ```
 
+### 3. 预测
+
+使用单个GPU通过如下命令预测一个视频，并保存为视频
+
+```bash
+# 预测一个视频
+CUDA_VISIBLE_DEVICES=0 python tools/infer_mot.py -c configs/mot/fairmot/fairmot_dla34_30e_1088x608.yml -o weights=https://paddledet.bj.bcebos.com/models/mot/fairmot_dla34_30e_1088x608.pdparams --video_file={your video name}.mp4  --save_videos
+```
+
+**注意:**
+ 请先确保已经安装了[ffmpeg](https://ffmpeg.org/ffmpeg.html), Linux(Ubuntu)平台可以直接用以下命令安装：`apt-get update && apt-get install -y ffmpeg`。
+
 ## 引用
 ```
 @article{wang2019towards,
diff --git a/configs/mot/deepsort/README.md b/configs/mot/deepsort/README.md
index 4847be341cf3c02306ed0f7793b47b6b0afc1192..237d3bd70e2ee1f83de898b77c1b41e9674fce29 100644
--- a/configs/mot/deepsort/README.md
+++ b/configs/mot/deepsort/README.md
@@ -53,7 +53,7 @@ Each txt is the detection result of all the pictures extracted from each video,
 CUDA_VISIBLE_DEVICES=0 python tools/eval_mot.py -c configs/mot/jde/jde_darknet53_30e_1088x608_track.yml -o weights=https://paddledet.bj.bcebos.com/models/mot/jde_darknet53_30e_1088x608.pdparams
 
 # use saved checkpoint after training
-CUDA_VISIBLE_DEVICES=0 python tools/eval_mot.py -c configs/mot/jde/jde_darknet53_30e_1088x608_track.yml -o weights=output/jde_darknet53_30e_1088x608/model_final
+CUDA_VISIBLE_DEVICES=0 python tools/eval_mot.py -c configs/mot/jde/jde_darknet53_30e_1088x608_track.yml -o weights=output/jde_darknet53_30e_1088x608/model_final.pdparams
 ```
 
 ### 2. Tracking
diff --git a/configs/mot/deepsort/README_cn.md b/configs/mot/deepsort/README_cn.md
index 81af5f6c1e0be964d4f51b8d9a0127992f366778..203df6a811e8453072b19f41ee36a1e9e2f4d84d 100644
--- a/configs/mot/deepsort/README_cn.md
+++ b/configs/mot/deepsort/README_cn.md
@@ -53,7 +53,7 @@ det_results_dir
 CUDA_VISIBLE_DEVICES=0 python tools/eval_mot.py -c configs/mot/jde/jde_darknet53_30e_1088x608_track.yml -o weights=https://paddledet.bj.bcebos.com/models/mot/jde_darknet53_30e_1088x608.pdparams
 
 # 使用训练保存的checkpoint
-CUDA_VISIBLE_DEVICES=0 python tools/eval_mot.py -c configs/mot/jde/jde_darknet53_30e_1088x608_track.yml -o weights=output/jde_darknet53_30e_1088x608/model_final
+CUDA_VISIBLE_DEVICES=0 python tools/eval_mot.py -c configs/mot/jde/jde_darknet53_30e_1088x608_track.yml -o weights=output/jde_darknet53_30e_1088x608/model_final.pdparams
 ```
 
 ### 2. 跟踪预测
diff --git a/configs/mot/fairmot/README.md b/configs/mot/fairmot/README.md
index 0254ef340e967135ec6cac1f7c6899b97c2a8a23..e6b7d74910cc4814d3fa4a76c359b0d7c9ebbb5d 100644
--- a/configs/mot/fairmot/README.md
+++ b/configs/mot/fairmot/README.md
@@ -53,8 +53,7 @@ Evaluating the track performance of FairMOT on val dataset in single GPU with fo
 CUDA_VISIBLE_DEVICES=0 python tools/eval_mot.py -c configs/mot/fairmot/fairmot_dla34_30e_1088x608.yml -o weights=https://paddledet.bj.bcebos.com/models/mot/fairmot_dla34_30e_1088x608.pdparams
 
 # use saved checkpoint in training
-CUDA_VISIBLE_DEVICES=0 python tools/eval_mot.py -c configs/mot/fairmot/fairmot_dla34_30e_1088x608.yml -o weights=output/fairmot_dla34_30e_1088x608/model_final
-```
+CUDA_VISIBLE_DEVICES=0 python tools/eval_mot.py -c configs/mot/fairmot/fairmot_dla34_30e_1088x608.yml -o weights=output/fairmot_dla34_30e_1088x608/model_final.pdparams
 
 ## Citations
 ```
diff --git a/configs/mot/fairmot/README_cn.md b/configs/mot/fairmot/README_cn.md
index 8bab81c78ae397cfbc17583cc3522de75bbba210..f6b4fac48dd032c67a6a1df57f9a5b17905fe70f 100644
--- a/configs/mot/fairmot/README_cn.md
+++ b/configs/mot/fairmot/README_cn.md
@@ -50,7 +50,7 @@ python -m paddle.distributed.launch --log_dir=./fairmot_dla34_30e_1088x608/ --gp
 CUDA_VISIBLE_DEVICES=0 python tools/eval_mot.py -c configs/mot/fairmot/fairmot_dla34_30e_1088x608.yml -o weights=https://paddledet.bj.bcebos.com/models/mot/fairmot_dla34_30e_1088x608.pdparams
 
 # 使用训练保存的checkpoint
-CUDA_VISIBLE_DEVICES=0 python tools/eval_mot.py -c configs/mot/fairmot/fairmot_dla34_30e_1088x608.yml -o weights=output/fairmot_dla34_30e_1088x608/model_final
+CUDA_VISIBLE_DEVICES=0 python tools/eval_mot.py -c configs/mot/fairmot/fairmot_dla34_30e_1088x608.yml -o weights=output/fairmot_dla34_30e_1088x608/model_final.pdparams
 ```
 
 ## 引用
diff --git a/configs/mot/fairmot/_base_/fairmot_reader_1088x608.yml b/configs/mot/fairmot/_base_/fairmot_reader_1088x608.yml
index 6dadac3a32f8fe46a06d73374ce147e553a21db2..5c0d4824a83cf0d82d6f2be59a99c63ca13d22e1 100644
--- a/configs/mot/fairmot/_base_/fairmot_reader_1088x608.yml
+++ b/configs/mot/fairmot/_base_/fairmot_reader_1088x608.yml
@@ -28,3 +28,14 @@ EvalMOTReader:
     - NormalizeImage: {mean: [0, 0, 0], std: [1, 1, 1]}
     - Permute: {to_rgb: True}
   batch_size: 1
+
+
+TestMOTReader:
+  inputs_def:
+    image_shape: [3, 608, 1088]
+  sample_transforms:
+    - Decode: {to_rgb: False}
+    - LetterBoxResize: {target_size: [608, 1088]}
+    - NormalizeImage: {mean: [0, 0, 0], std: [1, 1, 1]}
+    - Permute: {to_rgb: True}
+  batch_size: 1
diff --git a/configs/mot/jde/README.md b/configs/mot/jde/README.md
index d8ace647b9ebe97a9e288043bd0dada42d4e4a68..4e3893d1c6e4b1dac8fc77c78b6a9aeaa37d1d43 100644
--- a/configs/mot/jde/README.md
+++ b/configs/mot/jde/README.md
@@ -48,20 +48,9 @@ Evaluating the track performance of JDE on val dataset in single GPU with follow
 CUDA_VISIBLE_DEVICES=0 python tools/eval_mot.py -c configs/mot/jde/jde_darknet53_30e_1088x608.yml -o weights=https://paddledet.bj.bcebos.com/models/mot/jde_darknet53_30e_1088x608.pdparams
 
 # use saved checkpoint in training
-CUDA_VISIBLE_DEVICES=0 python tools/eval_mot.py -c configs/mot/jde/jde_darknet53_30e_1088x608.yml -o weights=output/jde_darknet53_30e_1088x608/model_final
+CUDA_VISIBLE_DEVICES=0 python tools/eval_mot.py -c configs/mot/jde/jde_darknet53_30e_1088x608.yml -o weights=output/jde_darknet53_30e_1088x608/model_final.pdparams
 ```
 
-### 3. Inference
-
-Inference a vidoe in single GPU with following commands.
-
-```bash
-# inference on video and save a video
-CUDA_VISIBLE_DEVICES=0 python tools/infer_mot.py -c configs/mot/jde/jde_darknet53_30e_1088x608.yml -o weights=https://paddledet.bj.bcebos.com/models/mot/jde_darknet53_30e_1088x608.pdparams --video_file={your video name}.mp4 --save_videos
-```
-**Notes:**
- Please make sure that [ffmpeg](https://www.ffmpeg.org) is installed first.
-
 ## Citations
 ```
 @article{wang2019towards,
diff --git a/configs/mot/jde/README_cn.md b/configs/mot/jde/README_cn.md
index 50672cf4e4c583e96876e507120ec88c90a3e63c..f2577268285b0c565e7163978a9c2ea84800b318 100644
--- a/configs/mot/jde/README_cn.md
+++ b/configs/mot/jde/README_cn.md
@@ -47,20 +47,9 @@ python -m paddle.distributed.launch --log_dir=./jde_darknet53_30e_1088x608/ --gp
 CUDA_VISIBLE_DEVICES=0 python tools/eval_mot.py -c configs/mot/jde/jde_darknet53_30e_1088x608.yml -o weights=https://paddledet.bj.bcebos.com/models/mot/jde_darknet53_30e_1088x608.pdparams
 
 # 使用训练保存的checkpoint
-CUDA_VISIBLE_DEVICES=0 python tools/eval_mot.py -c configs/mot/jde/jde_darknet53_30e_1088x608.yml -o weights=output/jde_darknet53_30e_1088x608/model_final
+CUDA_VISIBLE_DEVICES=0 python tools/eval_mot.py -c configs/mot/jde/jde_darknet53_30e_1088x608.yml -o weights=output/jde_darknet53_30e_1088x608/model_final.pdparams
 ```
 
-### 3. 预测
-
-使用单个GPU通过如下命令预测一个视频，并保存为视频
-
-```bash
-# 预测一个视频
-CUDA_VISIBLE_DEVICES=0 python tools/infer_mot.py -c configs/mot/jde/jde_darknet53_30e_1088x608.yml -o weights=https://paddledet.bj.bcebos.com/models/mot/jde_darknet53_30e_1088x608.pdparams --video_file={your video name}.mp4  --save_videos
-```
-**注意:**
- 请先确保已经安装了[ffmpeg](https://www.ffmpeg.org)。
-
 ## 引用
 ```
 @article{wang2019towards,
diff --git a/ppdet/engine/tracker.py b/ppdet/engine/tracker.py
index 153db303b640edf561ac18d1dffd5a6baa3ff3d0..4040d9200158c747249eaaf4d48a2cd88712d47b 100644
--- a/ppdet/engine/tracker.py
+++ b/ppdet/engine/tracker.py
@@ -280,8 +280,9 @@ class Tracker(object):
             timer_calls.append(tc)
 
             if save_videos:
-                output_video_path = os.path.join(save_dir, '{}.mp4'.format(seq))
-                cmd_str = 'ffmpeg -f image2 -i {}/%05d.jpg -c:v copy {}'.format(
+                output_video_path = os.path.join(save_dir, '..',
+                                                 '{}_vis.mp4'.format(seq))
+                cmd_str = 'ffmpeg -f image2 -i {}/%05d.jpg {}'.format(
                     save_dir, output_video_path)
                 os.system(cmd_str)
                 logger.info('Save video in {}.'.format(output_video_path))
@@ -367,8 +368,9 @@ class Tracker(object):
             raise ValueError(model_type)
 
         if save_videos:
-            output_video_path = os.path.join(save_dir, '{}.mp4'.format(seq))
-            cmd_str = 'ffmpeg -f image2 -i {}/%05d.jpg -c:v copy {}'.format(
+            output_video_path = os.path.join(save_dir, '..',
+                                             '{}_vis.mp4'.format(seq))
+            cmd_str = 'ffmpeg -f image2 -i {}/%05d.jpg {}'.format(
                 save_dir, output_video_path)
             os.system(cmd_str)
             logger.info('Save video in {}'.format(output_video_path))