Merge branch 'PaddlePaddle:develop' into develop

README_ch.md

@@ -25,13 +25,8 @@ PP-ShiTu图像识别系统效果展示
 - 🔥️ 2022.6.15 发布PULC超轻量图像分类方案，CPU推理3ms，精度比肩SwinTransformer，覆盖人、车、OCR场景九大常见任务。
 - 2022.5.26 [飞桨产业实践范例直播课](http://aglc.cn/v-c4FAR)，解读**超轻量重点区域人员出入管理方案**。
 - 2022.5.23 新增[人员出入管理范例库](https://aistudio.baidu.com/aistudio/projectdetail/4094475)，具体内容可以在 AI Stuio 上体验。
-- 2022.5.20 上线[PP-HGNet](./docs/zh_CN/models/PP-HGNet.md), [PP-LCNet v2](./docs/zh_CN/models/PP-LCNetV2.md)。
+- 2022.5.20 上线[PP-HGNet](./docs/zh_CN/models/PP-HGNet.md), [PP-LCNetv2](./docs/zh_CN/models/PP-LCNetV2.md)。
 - 2022.4.21 新增 CVPR2022 oral论文 [MixFormer](https://arxiv.org/pdf/2204.02557.pdf) 相关[代码](https://github.com/PaddlePaddle/PaddleClas/pull/1820/files)。
-- 2022.1.27 全面升级文档；新增[PaddleServing C++ pipeline部署方式](./deploy/paddleserving)和[18M图像识别安卓部署Demo](./deploy/lite_shitu)。
-- 2021.11.1 发布[PP-ShiTu技术报告](https://arxiv.org/pdf/2111.00775.pdf)，新增饮料识别demo。
-- 2021.10.23 发布轻量级图像识别系统PP-ShiTu，CPU上0.2s即可完成在10w+库的图像识别。
-[点击这里](./docs/zh_CN/quick_start/quick_start_recognition.md)立即体验。
-- 2021.09.17 发布PP-LCNet系列超轻量骨干网络模型, 在Intel CPU上，单张图像预测速度约5ms，ImageNet-1K数据集上Top1识别准确率达到80.82%，超越ResNet152的模型效果。PP-LCNet的介绍可以参考[论文](https://arxiv.org/pdf/2109.15099.pdf), 或者[PP-LCNet模型介绍](docs/zh_CN/models/PP-LCNet.md)，相关指标和预训练权重可以从 [这里](docs/zh_CN/algorithm_introduction/ImageNet_models.md)下载。
 - [more](./docs/zh_CN/others/update_history.md)
 
 ## 特性
@@ -51,16 +46,16 @@ PP-ShiTu图像识别系统效果展示
 
 ## 快速体验
 
-PULC超轻量图像分类方案快速体验：[点击这里](docs/zh_CN/PULC/PULC_person_exists.md)。
+PULC超轻量图像分类方案快速体验：[点击这里](docs/zh_CN/PULC/PULC_quickstart.md)
 
-PP-ShiTu图像识别快速体验：[点击这里](./docs/zh_CN/quick_start/quick_start_recognition.md)。
+PP-ShiTu图像识别快速体验：[点击这里](./docs/zh_CN/quick_start/quick_start_recognition.md)
 
 ## 文档教程
 - [环境准备](docs/zh_CN/installation/install_paddleclas.md)
 - PULC超轻量图像分类实用方案 文档更新中
-  - 超轻量图像分类模型库 文档更新中
+  - [超轻量图像分类模型库](docs/zh_CN/installation/PULC_model_list.md)
     - [PULC有人/无人分类模型](docs/zh_CN/PULC/PULC_person_exists.md)
-    - PULC人体属性识别模型 文档更新中
+    - [PULC人体属性识别模型](docs/zh_CN/PULC/PULC_person_attribute.md)
     - [PULC佩戴安全帽分类模型](docs/zh_CN/PULC/PULC_safety_helmet.md)
     - [PULC交通标志分类模型](docs/zh_CN/PULC/PULC_traffic_sign.md)
     - [PULC车辆属性识别模型](docs/zh_CN/PULC/PULC_vehicle_attribute.md)
@@ -92,7 +87,7 @@ PP-ShiTu图像识别快速体验：[点击这里](./docs/zh_CN/quick_start/quick
   - 模型压缩 文档更新中
 - PP系列骨干网络模型
   - [PP-HGNet](docs/zh_CN/models/PP-HGNet.md)
-  - [PP-LCNet v2](docs/zh_CN/models/PP-LCNetV2.md)
+  - [PP-LCNetv2](docs/zh_CN/models/PP-LCNetV2.md)
   - [PP-LCNet](docs/zh_CN/models/PP-LCNet.md)
 - [SSLD半监督知识蒸馏方案](docs/zh_CN/advanced_tutorials/ssld.md)
 - 前沿算法

deploy/configs/PULC/language_classification/inference_language_classification.yaml

@@ -28,6 +28,6 @@ PostProcess:
   main_indicator: Topk
   Topk:
     topk: 2
-    class_id_map_file: "../dataset/language_classification/label_list.txt"
+    class_id_map_file: "../ppcls/utils/PULC_label_list/language_classification_label_list.txt"
   SavePreLabel:
     save_dir: ./pre_label/

deploy/configs/PULC/text_image_orientation/inference_text_image_orientation.yaml

@@ -30,6 +30,6 @@ PostProcess:
   main_indicator: Topk
   Topk:
     topk: 2
-    class_id_map_file: "../dataset/text_image_orientation/label_list.txt"
+    class_id_map_file: "../ppcls/utils/PULC_label_list/text_image_orientation_label_list.txt"
   SavePreLabel:
     save_dir: ./pre_label/

deploy/configs/PULC/textline_orientation/inference_textline_orientation.yaml

@@ -28,6 +28,6 @@ PostProcess:
   main_indicator: Topk
   Topk:
     topk: 1
-    class_id_map_file: "../ppcls/utils/PULC/textline_orientation_label_list.txt"
+    class_id_map_file: "../ppcls/utils/PULC_label_list/textline_orientation_label_list.txt"
   SavePreLabel:
     save_dir: ./pre_label/

deploy/configs/PULC/vehicle_exists/inference_vehicle_exists.yaml

+Global:
+  infer_imgs: "./images/PULC/vehicle_exists/objects365_00001507.jpeg"
+  inference_model_dir: "./models/vehicle_exists_infer"
+  batch_size: 1
+  use_gpu: True
+  enable_mkldnn: False
+  cpu_num_threads: 10
+  enable_benchmark: True
+  use_fp16: False
+  ir_optim: True
+  use_tensorrt: False
+  gpu_mem: 8000
+  enable_profile: False
+
+PreProcess:
+  transform_ops:
+    - ResizeImage:
+        resize_short: 256
+    - CropImage:
+        size: 224
+    - NormalizeImage:
+        scale: 0.00392157
+        mean: [0.485, 0.456, 0.406]
+        std: [0.229, 0.224, 0.225]
+        order: ''
+        channel_num: 3
+    - ToCHWImage:
+
+PostProcess:
+  main_indicator: ThreshOutput
+  ThreshOutput:
+    threshold: 0.5
+    label_0: no_vehicle
+    label_1: contains_vehicle
+  SavePreLabel:
+    save_dir: ./pre_label/

docs/zh_CN/PULC/PULC_language_classification.md

-# PULC语种分类模型
+# PULC 语种分类模型
 
 ## 目录
 
-- [1.  模型和应用场景介绍](#1)
-- [2.  模型快速体验](#2)
+- [1. 模型和应用场景介绍](#1)
+- [2. 模型快速体验](#2)
   - [2.1 安装 paddleclas](#2.1)  
   - [2.2 预测](#2.2)
-- [3.  模型训练、评估和预测](#3)
+- [3. 模型训练、评估和预测](#3)
   - [3.1 环境配置](#3.1)  
   - [3.2 数据准备](#3.2)
     - [3.2.1 数据集来源](#3.2.1)
@@ -73,14 +73,14 @@ pip3 install paddleclas
 - 使用命令行快速预测
 
 ```
-paddleclas --model_name=language_classification --infer_imgs=deploy/images/PULC/language_classification/img_rot0_demo.jpg
+paddleclas --model_name=language_classification --infer_imgs=deploy/images/PULC/language_classification/word_35404.png
 ```
 
 结果如下：
 
 ```
 >>> result
-class_ids: [4, 9], scores: [0.96809, 0.01001], label_names: ['japan', 'latin'], filename: deploy/images/PULC/language_classification/word_35404.png
+class_ids: [4, 6], scores: [0.88672, 0.01434], label_names: ['japan', 'korean'], filename: deploy/images/PULC/language_classification/word_35404.png
 Predict complete!
 ```
 
@@ -99,7 +99,7 @@ print(next(result))
 
 ```
 >>> result
-[{'class_ids': [4, 9], 'scores': [0.96809, 0.01001], 'label_names': ['japan', 'latin'], 'filename': 'deploy/images/PULC/language_classification/word_35404.png'}]
+[{'class_ids': [4, 6], 'scores': [0.88672, 0.01434], 'label_names': ['japan', 'korean'], 'filename': '/deploy/images/PULC/language_classification/word_35404.png'}]
 ```
 
 <a name="3"></a>

docs/zh_CN/PULC/PULC_model_list.md

+# PULC 模型库
+
+------
+
+此处提供了 PULC 模型库的相关指标和模型的下载链接，其中预训练模型可以用来微调训练，推理模型可以直接用来预测和部署。
+
+
+|模型名称|模型简介|模型精度 |模型大小|推理耗时|下载地址|
+| --- | --- | --- | --- | --- | --- |
+| person_exists |[PULC有人/无人分类模型](PULC_person_exists.md)| 95.60 |6.5M|2.58ms|[推理模型](https://paddleclas.bj.bcebos.com/models/PULC/inference/person_exists_infer.tar) / [预训练模型](https://paddleclas.bj.bcebos.com/models/PULC/pretrained/person_exists_pretrained.pdparams)|
+| person_attribute |[PULC人体属性识别模型](PULC_person_attribute.md)| 78.59 |6.6M|2.01ms|[推理模型](https://paddleclas.bj.bcebos.com/models/PULC/inference/person_attribute_infer.tar) / [预训练模型](https://paddleclas.bj.bcebos.com/models/PULC/pretrained/person_attribute_pretrained.pdparams)|
+| safety_helmet |[PULC佩戴安全帽分类模型](PULC_safety_helmet.md)| 99.38 |6.5M|2.03ms|[推理模型](https://paddleclas.bj.bcebos.com/models/PULC/inference/safety_helmet_infer.tar) / [预训练模型](https://paddleclas.bj.bcebos.com/models/PULC/pretrained/safety_helmet_pretrained.pdparams)|
+| traffic_sign |[PULC交通标志分类模型](PULC_traffic_sign.md)| 98.35 |8.2M|2.10ms|[推理模型](https://paddleclas.bj.bcebos.com/models/PULC/inference/traffic_sign_infer.tar) / [预训练模型](https://paddleclas.bj.bcebos.com/models/PULC/pretrained/traffic_sign_pretrained.pdparams)|
+| vehicle_attribute |[PULC车辆属性识别模型](PULC_vehicle_attribute.md)| 90.81 |7.2M|2.36ms|[推理模型](https://paddleclas.bj.bcebos.com/models/PULC/inference/vehicle_attribute_infer.tar) / [预训练模型](https://paddleclas.bj.bcebos.com/models/PULC/pretrained/vehicle_attribute_pretrained.pdparams)|
+| vehicle_exists |PULC有车/无车分类模型 | 95.72 | 6.6M | 2.38ms |[推理模型](https://paddleclas.bj.bcebos.com/models/PULC/inference/vehicle_exists_infer.tar) / [预训练模型](https://paddleclas.bj.bcebos.com/models/PULC/pretrained/vehicle_exists_pretrained.pdparams)|
+| text_image_orientation |[PULC含文字图像方向分类模型](PULC_text_image_orientation.md)| 99.06 | 6.5M | 2.16ms |[推理模型](https://paddleclas.bj.bcebos.com/models/PULC/inference/text_image_orientation_infer.tar) / [预训练模型](https://paddleclas.bj.bcebos.com/models/PULC/pretrained/text_image_orientation_pretrained.pdparams)|
+| textline_orientation |[PULC文本行方向分类模型](PULC_textline_orientation.md)| 96.01 |6.5M|2.72ms|[推理模型](https://paddleclas.bj.bcebos.com/models/PULC/inference/textline_orientation_infer.tar) / [预训练模型](https://paddleclas.bj.bcebos.com/models/PULC/pretrained/textline_orientation_pretrained.pdparams)|
+| language_classification |[PULC语种分类模型](PULC_language_classification.md)| 99.26 |6.5M|2.58ms|[推理模型](https://paddleclas.bj.bcebos.com/models/PULC/inference/language_classification_infer.tar) / [预训练模型](https://paddleclas.bj.bcebos.com/models/PULC/pretrained/language_classification_pretrained.pdparams)|
+
+
+**备注：**
+
+* 以上所有的模型的 backbone 均为 PPLCNet_x1_0，部分模型大小不同是由于分类的输出大小不同导致的，推理耗时是基于Intel(R) Xeon(R) Gold 6148 CPU @ 2.40GHz 测试得到，其中测试过程开启 MKLDNN 加速策略，线程数为10。速度测试过程会有轻微波动。
+
+* person_exists、safety_helmet、vehicle_exists 的评测指标为 TprAtFpr，person_attribute、vehicle_attribute的评测指标为ma、traffic_sign、text_image_orientation、textline_orientation、language_classification的评测指标为Top-1 Acc。
\ No newline at end of file

docs/zh_CN/PULC/PULC_quickstart.md

+# PULC 快速体验
+
+------
+
+本文主要介绍PaddleClas whl包对 PULC 系列模型的快速使用。
+
+## 目录
+
+- [1. 安装](#1)
+  - [1.1 安装PaddlePaddle](#11)
+  - [1.2 安装PaddleClas whl包](#12)
+- [2. 快速体验](#2)
+  - [2.1 命令行使用](#2.1)
+  - [2.2 Python脚本使用](#2.2)
+- [3.小结](#3)
+
+<a name="1"></a>
+
+## 1. 安装
+
+<a name="1.1"></a>
+
+### 1.1 安装 PaddlePaddle
+
+- 您的机器安装的是 CUDA9 或 CUDA10，请运行以下命令安装
+
+  ```bash
+  python3 -m pip install paddlepaddle-gpu -i https://mirror.baidu.com/pypi/simple
+  ```
+
+- 您的机器是CPU，请运行以下命令安装
+
+  ```bash
+  python3 -m pip install paddlepaddle -i https://mirror.baidu.com/pypi/simple
+  ```
+
+更多的版本需求，请参照[飞桨官网安装文档](https://www.paddlepaddle.org.cn/install/quick)中的说明进行操作。
+
+<a name="1.2"></a>
+
+### 1.2 安装 PaddleClas whl 包
+
+```bash
+pip install paddleclas
+```
+
+<a name="2"></a>
+
+## 2. 快速体验
+
+PaddleClas 提供了一系列测试图片，里边包含人、车、OCR等方向的多个场景大的demo数据。点击[这里](https://paddleclas.bj.bcebos.com/data/PULC/pulc_demo_imgs.zip)下载并解压，然后在终端中切换到相应目录。
+
+<a name="2.1"></a>
+
+### 2.1 命令行使用
+
+```
+cd /path/to/pulc_demo_imgs
+```
+
+使用命令行预测：
+
+```bash
+paddleclas --model_name=person_exists --infer_imgs=pulc_demo_imgs/person_exists/objects365_01780782.jpg
+```
+
+结果如下：
+```
+>>> result
+class_ids: [0], scores: [0.9955421453341842], label_names: ['nobody'], filename: pulc_demo_imgs/person_exists/objects365_01780782.jpg
+Predict complete!
+```
+
+若预测结果为 `nobody`，表示该图中没有人，若预测结果为 `someone`，则表示该图中有人。此处预测结果为 `nobody`，表示该图中没有人。
+
+**备注**： 更换其他预测的数据时，只需要改变 `--infer_imgs=xx` 中的字段即可，支持传入整个文件夹，如需要替换模型，更改 `--model_name` 中的模型名字即可，模型名字可以参考[模型库](./PULC_model_list.md)。
+
+<a name="2.2"></a>
+
+### 2.2 Python 脚本使用
+
+此处提供了在 python 脚本中使用 PULC 有人/无人分类模型预测的例子。
+
+```python
+import paddleclas
+model = paddleclas.PaddleClas(model_name="person_exists")
+result = model.predict(input_data="pulc_demo_imgs/person_exists/objects365_01780782.jpg")
+print(next(result))
+```
+
+打印的结果如下：
+
+```
+>>> result
+[{'class_ids': [0], 'scores': [0.9955421453341842], 'label_names': ['nobody'], 'filename': 'pulc_demo_imgs/person_exists/objects365_01780782.jpg'}]
+```
+
+**备注**：`model.predict()` 为可迭代对象（`generator`），因此需要使用 `next()` 函数或 `for` 循环对其迭代调用。每次调用将以 `batch_size` 为单位进行一次预测，并返回预测结果, 默认 `batch_size` 为 1，如果需要更改 `batch_size`，实例化模型时，需要指定 `batch_size`，如 `model = paddleclas.PaddleClas(model_name="person_exists",  batch_size=2)`。更换其他模型只需要替换`model_name`, `model_name`,可以参考[模型库](./PULC_model_list.md)。
+
+
+<a name="3"></a>
+
+## 3. 小结
+
+通过本节内容，相信您已经熟练掌握 PaddleClas whl 包的 PULC 模型使用方法并获得了初步效果。
+
+PULC 方法产出的系列模型在人、车、OCR等方向的多个场景中均验证有效，用超轻量模型就可实现与 SwinTransformer 模型接近的精度，预测速度提高 40+ 倍。并且打通数据、模型训练、压缩和推理部署全流程，您可以参考[文档教程](./PULC_train.md)，正式开启 PULC 的体验之旅。
+
+
+
+

docs/zh_CN/PULC/PULC_text_image_orientation.md

-# PULC含文字图像方向分类模型
+# PULC 含文字图像方向分类模型
 
 ## 目录
 
-- [1.  模型和应用场景介绍](#1)
-- [2.  模型快速体验](#2)
+- [1. 模型和应用场景介绍](#1)
+- [2. 模型快速体验](#2)
   - [2.1 安装 paddleclas](#2.1)  
   - [2.2 预测](#2.2)
-
-- [3.  模型训练、评估和预测](#3)
+- [3. 模型训练、评估和预测](#3)
   - [3.1 环境配置](#3.1)  
   - [3.2 数据准备](#3.2)
     - [3.2.1 数据集来源](#3.2.1)

docs/zh_CN/PULC/PULC_traffic_sign.md

@@ -78,13 +78,13 @@ pip3 install paddlepaddle paddleclas
 * 使用命令行快速预测
 
 ```bash
-paddleclas --model_name traffic_sign  --infer_imgs PaddleClas/deploy/images/PULC/traffic_sign/100999_83928.jpg
+paddleclas --model_name traffic_sign  --infer_imgs deploy/images/PULC/traffic_sign/100999_83928.jpg
 ```
 
 结果如下：
 ```
 >>> result
-class_ids: [182, 179, 162, 128, 24], scores: [0.98623, 0.01255, 0.00022, 0.00021, 0.00012], label_names: ['pl110', 'pl100', 'pl120', 'p26', 'pm10'], filename: PaddleClas/deploy/images/PULC/traffic_sign/100999_83928.jpg
+class_ids: [182, 179, 162, 128, 24], scores: [0.98623, 0.01255, 0.00022, 0.00021, 0.00012], label_names: ['pl110', 'pl100', 'pl120', 'p26', 'pm10'], filename: deploy/images/PULC/traffic_sign/100999_83928.jpg
 ```
 
 **备注**： 更换其他预测的数据时，只需要改变 `--infer_imgs=xx` 中的字段即可，支持传入整个文件夹。
@@ -94,7 +94,7 @@ class_ids: [182, 179, 162, 128, 24], scores: [0.98623, 0.01255, 0.00022, 0.00021
 ```python
 import paddleclas
 model = paddleclas.PaddleClas(model_name="traffic_sign")
-result = model.predict(input_data="PaddleClas/deploy/images/PULC/traffic_sign/100999_83928.jpg")
+result = model.predict(input_data="deploy/images/PULC/traffic_sign/100999_83928.jpg")
 print(next(result))
 ```
 
@@ -102,7 +102,7 @@ print(next(result))
 
 ```
 result
-[{'class_ids': [182, 179, 162, 128, 24], 'scores': [0.98623, 0.01255, 0.00022, 0.00021, 0.00012], 'label_names': ['pl110', 'pl100', 'pl120', 'p26', 'pm10'], 'filename': 'PaddleClas/deploy/images/PULC/traffic_sign/100999_83928.jpg'}]
+[{'class_ids': [182, 179, 162, 128, 24], 'scores': [0.98623, 0.01255, 0.00022, 0.00021, 0.00012], 'label_names': ['pl110', 'pl100', 'pl120', 'p26', 'pm10'], 'filename': 'deploy/images/PULC/traffic_sign/100999_83928.jpg'}]
 ```
 
 <a name="3"></a>

docs/zh_CN/PULC/PULC_vehicle_attribute.md

@@ -47,9 +47,9 @@
 | Res2Net200_vd_26w_4s  | 91.36 | 79.46  | 293 | 使用ImageNet预训练模型 |
 | ResNet50  | 89.98 | 12.83  | 92 | 使用ImageNet预训练模型 |
 | MobileNetV3_small_x0_35  | 87.41 | 2.91  | 2.8 | 使用ImageNet预训练模型 |
-| PPLCNet_x1_0  | 89.57 | 2.36  | 8.2 | 使用ImageNet预训练模型 |
-| PPLCNet_x1_0  | 90.07 | 2.36  | 8.2 | 使用SSLD预训练模型 |
-| PPLCNet_x1_0  | 90.59 | 2.36  | 8.2 | 使用SSLD预训练模型+EDA策略|
+| PPLCNet_x1_0  | 89.57 | 2.36  | 7.2 | 使用ImageNet预训练模型 |
+| PPLCNet_x1_0  | 90.07 | 2.36  | 7.2 | 使用SSLD预训练模型 |
+| PPLCNet_x1_0  | 90.59 | 2.36  | 7.2 | 使用SSLD预训练模型+EDA策略|
 | <b>PPLCNet_x1_0<b>  | <b>90.81<b> | <b>2.36<b>  | <b>8.2<b> | 使用SSLD预训练模型+EDA策略+SKL-UGI知识蒸馏策略|
 
 从表中可以看出，backbone 为 Res2Net200_vd_26w_4s 时精度较高，但是推理速度较慢。将 backbone 替换为轻量级模型 MobileNetV3_small_x0_35 后，速度可以大幅提升，但是精度下降明显。将 backbone 替换为 PPLCNet_x1_0 时，精度提升 2.16%，同时速度也提升 23% 左右。在此基础上，使用 SSLD 预训练模型后，在不改变推理速度的前提下，精度可以提升约 0.5%，进一步地，当融合EDA策略后，精度可以再提升 0.52%，最后，在使用 SKL-UGI 知识蒸馏后，精度可以继续提升 0.23%。此时，PPLCNet_x1_0 的精度与 Res2Net200_vd_26w_4s 仅相差0.55%，但是速度快32倍。关于 PULC 的训练方法和推理部署方法将在下面详细介绍。
@@ -79,13 +79,13 @@ pip3 install paddlepaddle paddleclas
 * 使用命令行快速预测
 
 ```bash
-paddleclas --model_name vehicle_attribute --infer_imgs PaddleClas/deploy/images/PULC/vehicle_attribute/0002_c002_00030670_0.jpg
+paddleclas --model_name vehicle_attribute --infer_imgs deploy/images/PULC/vehicle_attribute/0002_c002_00030670_0.jpg
 ```
 
 结果如下：
 ```
 >>> result
-attributes: Color: (yellow, prob: 0.9893476963043213), Type: (hatchback, prob: 0.9734097719192505), output: [1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0], filename: PaddleClas/deploy/images/PULC/vehicle_attribute/0002_c002_00030670_0.jpg
+attributes: Color: (yellow, prob: 0.9893476963043213), Type: (hatchback, prob: 0.9734097719192505), output: [1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0], filename: deploy/images/PULC/vehicle_attribute/0002_c002_00030670_0.jpg
 ppcls INFO: Predict complete!
 ```
 
@@ -96,7 +96,7 @@ ppcls INFO: Predict complete!
 ```python
 import paddleclas
 model = paddleclas.PaddleClas(model_name="vehicle_attribute")
-result = model.predict(input_data="PaddleClas/deploy/images/PULC/vehicle_attribute/0002_c002_00030670_0.jpg")
+result = model.predict(input_data="deploy/images/PULC/vehicle_attribute/0002_c002_00030670_0.jpg")
 print(next(result))
 ```
 
@@ -104,7 +104,7 @@ print(next(result))
 
 ```
 result
-[{'attributes': 'Color: (yellow, prob: 0.9893476963043213), Type: (hatchback, prob: 0.9734097719192505)', 'output': [1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0], 'filename': 'PaddleClas/deploy/images/PULC/vehicle_attribute/0002_c002_00030670_0.jpg'}]
+[{'attributes': 'Color: (yellow, prob: 0.9893476963043213), Type: (hatchback, prob: 0.9734097719192505)', 'output': [1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0], 'filename': 'deploy/images/PULC/vehicle_attribute/0002_c002_00030670_0.jpg'}]
 ```
 
 <a name="3"></a>

paddleclas.py

@@ -178,7 +178,7 @@ IMN_MODEL_SERIES = {
 PULC_MODEL_BASE_DOWNLOAD_URL = "https://paddleclas.bj.bcebos.com/models/PULC/{}_infer.tar"
 PULC_MODELS = [
     "person_exists", "person_attribute", "safety_helmet", "traffic_sign",
-    "vehicle_exists", "vehicle_attr", "textline_orientation",
+    "vehicle_exists", "vehicle_attribute", "textline_orientation",
     "text_image_orientation", "language_classification"
 ]
 
@@ -247,8 +247,10 @@ def init_config(model_type, model_name, inference_model_dir, **kwargs):
             cfg.PostProcess.Topk.class_id_map_file = kwargs[
                 "class_id_map_file"]
         else:
-            cfg.PostProcess.Topk.class_id_map_file = os.path.relpath(
+            class_id_map_file_path = os.path.relpath(
                 cfg.PostProcess.Topk.class_id_map_file, "../")
+            cfg.PostProcess.Topk.class_id_map_file = os.path.join(
+                __dir__, class_id_map_file_path)
     if "VehicleAttribute" in cfg.PostProcess:
         if "color_threshold" in kwargs and kwargs["color_threshold"]:
             cfg.PostProcess.VehicleAttribute.color_threshold = kwargs[

ppcls/configs/PULC/language_classification/MobileNetV3_small_x0_35.yaml

@@ -121,7 +121,7 @@ Infer:
   PostProcess:
     name: Topk
     topk: 2
-    class_id_map_file: ppcls/utils/PULC/language_classification_label_list.txt
+    class_id_map_file: ppcls/utils/PULC_label_list/language_classification_label_list.txt
 
 Metric:
   Train:

ppcls/configs/PULC/language_classification/PPLCNet_x1_0.yaml

@@ -132,7 +132,7 @@ Infer:
   PostProcess:
     name: Topk
     topk: 2
-    class_id_map_file: ppcls/utils/PULC/language_classification_label_list.txt
+    class_id_map_file: ppcls/utils/PULC_label_list/language_classification_label_list.txt
 
 Metric:
   Train:

ppcls/configs/PULC/language_classification/PPLCNet_x1_0_distillation.yaml

@@ -152,7 +152,7 @@ Infer:
   PostProcess:
     name: Topk
     topk: 2
-    class_id_map_file: ppcls/utils/PULC/language_classification_label_list.txt
+    class_id_map_file: ppcls/utils/PULC_label_list/language_classification_label_list.txt
 
 Metric:
     Train:
@@ -161,4 +161,4 @@ Metric:
         topk: [1, 2]
     Eval:
     - TopkAcc:
-        topk: [1, 2]
\ No newline at end of file
+        topk: [1, 2]

ppcls/configs/PULC/language_classification/PPLCNet_x1_0_search.yaml

@@ -131,7 +131,7 @@ Infer:
   PostProcess:
     name: Topk
     topk: 2
-    class_id_map_file: ppcls/utils/PULC/language_classification_label_list.txt
+    class_id_map_file: ppcls/utils/PULC_label_list/language_classification_label_list.txt
 
 Metric:
   Train:

ppcls/configs/PULC/language_classification/SwinTransformer_tiny_patch4_window7_224.yaml

@@ -152,7 +152,7 @@ Infer:
   PostProcess:
     name: Topk
     topk: 2
-    class_id_map_file: ppcls/utils/PULC/language_classification_label_list.txt
+    class_id_map_file: ppcls/utils/PULC_label_list/language_classification_label_list.txt
 
 Metric:
   Eval:

ppcls/configs/PULC/text_image_orientation/MobileNetV3_small_x0_35.yaml

@@ -121,7 +121,7 @@ Infer:
   PostProcess:
     name: Topk
     topk: 2
-    class_id_map_file: ppcls/utils/PULC/text_image_orientation_label_list.txt
+    class_id_map_file: ppcls/utils/PULC_label_list/text_image_orientation_label_list.txt
 
 Metric:
   Train:

ppcls/configs/PULC/text_image_orientation/PPLCNet_x1_0.yaml

@@ -132,7 +132,7 @@ Infer:
   PostProcess:
     name: Topk
     topk: 2
-    class_id_map_file: ppcls/utils/PULC/text_image_orientation_label_list.txt
+    class_id_map_file: ppcls/utils/PULC_label_list/text_image_orientation_label_list.txt
 
 Metric:
   Train:

ppcls/configs/PULC/text_image_orientation/PPLCNet_x1_0_distillation.yaml

@@ -152,7 +152,7 @@ Infer:
   PostProcess:
     name: Topk
     topk: 2
-    class_id_map_file: ppcls/utils/PULC/text_image_orientation_label_list.txt
+    class_id_map_file: ppcls/utils/PULC_label_list/text_image_orientation_label_list.txt
 
 Metric:
     Train:
@@ -161,4 +161,4 @@ Metric:
         topk: [1, 2]
     Eval:
     - TopkAcc:
-        topk: [1, 2]
\ No newline at end of file
+        topk: [1, 2]

ppcls/configs/PULC/text_image_orientation/PPLCNet_x1_0_search.yaml

@@ -135,7 +135,7 @@ Infer:
   PostProcess:
     name: Topk
     topk: 2
-    class_id_map_file: ppcls/utils/PULC/text_image_orientation_label_list.txt
+    class_id_map_file: ppcls/utils/PULC_label_list/text_image_orientation_label_list.txt
 
 Metric:
   Train:

ppcls/configs/PULC/text_image_orientation/SwinTransformer_tiny_patch4_window7_224.yaml

@@ -149,7 +149,7 @@ Infer:
   PostProcess:
     name: Topk
     topk: 2
-    class_id_map_file: ppcls/utils/PULC/text_image_orientation_label_list.txt
+    class_id_map_file: ppcls/utils/PULC_label_list/text_image_orientation_label_list.txt
 
 Metric:
   Eval:

ppcls/configs/PULC/textline_orientation/MobileNetV3_small_x0_35.yaml

@@ -123,7 +123,7 @@ Infer:
   PostProcess:
     name: Topk
     topk: 1
-    class_id_map_file: ppcls/utils/PULC/textline_orientation_label_list.txt
+    class_id_map_file: ppcls/utils/PULC_label_list/textline_orientation_label_list.txt
 
 Metric:
   Train:

ppcls/configs/PULC/textline_orientation/PPLCNet_x1_0.yaml

@@ -132,7 +132,7 @@ Infer:
   PostProcess:
     name: Topk
     topk: 1
-    class_id_map_file: ppcls/utils/PULC/textline_orientation_label_list.txt
+    class_id_map_file: ppcls/utils/PULC_label_list/textline_orientation_label_list.txt
 
 Metric:
   Train:

ppcls/configs/PULC/textline_orientation/PPLCNet_x1_0_224x224.yaml

@@ -121,7 +121,7 @@ Infer:
   PostProcess:
     name: Topk
     topk: 1
-    class_id_map_file: ppcls/utils/PULC/textline_orientation_label_list.txt
+    class_id_map_file: ppcls/utils/PULC_label_list/textline_orientation_label_list.txt
 
 Metric:
   Train:

ppcls/configs/PULC/textline_orientation/PPLCNet_x1_0_distillation.yaml

@@ -150,7 +150,7 @@ Infer:
   PostProcess:
     name: Topk
     topk: 1
-    class_id_map_file: ppcls/utils/PULC/textline_orientation_label_list.txt
+    class_id_map_file: ppcls/utils/PULC_label_list/textline_orientation_label_list.txt
 
 Metric:
     Train:

ppcls/configs/PULC/textline_orientation/PPLCNet_x1_0_search.yaml

@@ -133,7 +133,7 @@ Infer:
   PostProcess:
     name: Topk
     topk: 1
-    class_id_map_file: ppcls/utils/PULC/textline_orientation_label_list.txt 
+    class_id_map_file: ppcls/utils/PULC_label_list/textline_orientation_label_list.txt 
 
 Metric:
   Train:

ppcls/configs/PULC/textline_orientation/SwinTransformer_tiny_patch4_window7_224.yaml

@@ -153,7 +153,7 @@ Infer:
   PostProcess:
     name: Topk
     topk: 1
-    class_id_map_file: ppcls/utils/PULC/textline_orientation_label_list.txt
+    class_id_map_file: ppcls/utils/PULC_label_list/textline_orientation_label_list.txt
 
 Metric:
   Train:

ppcls/configs/PULC/traffic_sign/MobileNetV3_samll_x0_35.yaml

@@ -120,7 +120,7 @@ Infer:
   PostProcess:
     name: Topk
     topk: 5
-    class_id_map_file: dataset/traffic_sign/label_name_id.txt
+    class_id_map_file: ppcls/utils/PULC_label_list/traffic_sign_label_list.txt
 
 Metric:
   Train:

ppcls/configs/PULC/traffic_sign/PPLCNet_x1_0.yaml

@@ -137,7 +137,7 @@ Infer:
   PostProcess:
     name: Topk
     topk: 5
-    class_id_map_file: dataset/traffic_sign/label_name_id.txt
+    class_id_map_file: ppcls/utils/PULC_label_list/traffic_sign_label_list.txt
 
 Metric:
   Train:

ppcls/configs/PULC/traffic_sign/PPLCNet_x1_0_distillation.yaml

@@ -160,7 +160,7 @@ Infer:
   PostProcess:
     name: Topk
     topk: 5
-    class_id_map_file: dataset/traffic_sign/label_name_id.txt
+    class_id_map_file: ppcls/utils/PULC_label_list/traffic_sign_label_list.txt
 
 Metric:
     Train:

ppcls/configs/PULC/traffic_sign/PPLCNet_x1_0_search.yaml

@@ -138,7 +138,7 @@ Infer:
   PostProcess:
     name: Topk
     topk: 5
-    class_id_map_file: dataset/traffic_sign/label_name_id.txt
+    class_id_map_file: ppcls/utils/PULC_label_list/traffic_sign_label_list.txt
 
 Metric:
   Train:

ppcls/configs/PULC/traffic_sign/SwinTransformer_tiny_patch4_window7_224.yaml

@@ -157,7 +157,7 @@ Infer:
   PostProcess:
     name: Topk
     topk: 5
-    class_id_map_file: dataset/traffic_sign/label_name_id.txt
+    class_id_map_file: ppcls/utils/PULC_label_list/traffic_sign_label_list.txt
 
 Metric:
   Train:

ppcls/utils/PULC/text_image_orientation_label_list.txt

-0 0
-1 90
-2 180
-3 270