- 1. Introduction
- 2. Quick Start
- 3. Training, Evaluation and Inference
- 4. Model Compression
- 5. SHAS
- 6. Inference Deployment
This case provides a way for users to quickly build a lightweight, high-precision and practical classification model of traffic sign using PaddleClas PULC (Practical Ultra Lightweight image Classification). The model can be widely used in automatic driving, road monitoring, etc.
The following table lists the relevant indicators of the model. The first two lines means that using SwinTransformer_tiny and MobileNetV3_small_x0_35 as the backbone to training. The third to sixth lines means that the backbone is replaced by PPLCNet, additional use of EDA strategy and additional use of EDA strategy and SKL-UGI knowledge distillation strategy.
| Backbone | Top-1 Acc(%) | Latency(ms) | Size(M) | Training Strategy |
|---|---|---|---|---|
| SwinTranformer_tiny | 98.11 | 89.45 | 111 | using ImageNet pretrained model |
| MobileNetV3_small_x0_35 | 93.88 | 3.01 | 3.9 | using ImageNet pretrained model |
| PPLCNet_x1_0 | 97.78 | 2.10 | 8.2 | using ImageNet pretrained model |
| PPLCNet_x1_0 | 97.84 | 2.10 | 8.2 | using SSLD pretrained model |
| PPLCNet_x1_0 | 98.14 | 2.10 | 8.2 | using SSLD pretrained model + EDA strategy |
| PPLCNet_x1_0 | 98.35 | 2.10 | 8.2 | using SSLD pretrained model + EDA strategy + SKL-UGI knowledge distillation strategy |
It can be seen that high accuracy can be getted when backbone is SwinTranformer_tiny, but the speed is slow. Replacing backbone with the lightweight model MobileNetV3_small_x0_35, the speed can be greatly improved, but the accuracy will be greatly reduced. Replacing backbone with faster backbone PPLCNet_x1_0, the accuracy is lower 3.9 percentage points than MobileNetv3_small_x0_35. At the same time, the speed can be more than 43% faster. After additional using the SSLD pretrained model, the accuracy can be improved by about 0.06 percentage points without affecting the inference speed. Further, additional using the EDA strategy, the accuracy can be increased by 0.3 percentage points. Finally, after additional using the SKL-UGI knowledge distillation, the accuracy can be further improved by 0.21 percentage points. At this point, the accuracy exceeds that of SwinTranformer_tiny, but the speed is more than 41 times faster. The training method and deployment instructions of PULC will be introduced in detail below.
Note:
- The Latency is tested on Intel(R) Xeon(R) Gold 6148 CPU @ 2.40GHz. The MKLDNN is enabled and the number of threads is 10.
- About PP-LCNet, please refer to PP-LCNet Introduction and PP-LCNet Paper.
- Run the following command to install if CUDA9 or CUDA10 is available.
python3 -m pip install paddlepaddle-gpu -i https://mirror.baidu.com/pypi/simple- Run the following command to install if GPU device is unavailable.
python3 -m pip install paddlepaddle -i https://mirror.baidu.com/pypi/simplePlease refer to PaddlePaddle Installation for more information about installation, for examples other versions.
The command of PaddleClas installation as bellow:
pip3 install paddleclasFirst, please click here to download and unzip to get the test demo images.
- Prediction with CLI
paddleclas --model_name=traffic_sign --infer_imgs=pulc_demo_imgs/traffic_sign/100999_83928.jpgResults:
>>> 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: pulc_demo_imgs/traffic_sign/100999_83928.jpg
Predict complete!
Note: If you want to test other images, only need to specify the --infer_imgs argument, and the directory containing images is also supported.
- Prediction in Python
import paddleclas
model = paddleclas.PaddleClas(model_name="traffic_sign")
result = model.predict(input_data="pulc_demo_imgs/traffic_sign/100999_83928.jpg")
print(next(result))Note: The result returned by model.predict() is a generator, so you need to use the next() function to call it or for loop to loop it. And it will predict with batch_size size batch and return the prediction results when called. The default batch_size is 1, and you also specify the batch_size when instantiating, such as model = paddleclas.PaddleClas(model_name="traffic_sign", batch_size=2). The result of demo above:
>>> 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': 'pulc_demo_imgs/traffic_sign/100999_83928.jpg'}]
Please refer to Installation to get the description about installation.
All datasets used in this case are open source data. Train data is the subset of MS-COCO training data. And the validation data is the subset of Object365 training data. ImageNet_val is ImageNet-1k validation data.
The dataset used in this case is based on the Tsinghua-Tencent 100K dataset (CC-BY-NC license), TT100K randomly expanded and cropped according to the bounding box.
The processing to TT00K includes randomly expansion and cropping, details are shown below.
def get_random_crop_box(xmin, ymin, xmax, ymax, img_height, img_width, ratio=1.0):
h = ymax - ymin
w = ymax - ymin
xmin_diff = random.random() * ratio * min(w, xmin/ratio)
ymin_diff = random.random() * ratio * min(h, ymin/ratio)
xmax_diff = random.random() * ratio * min(w, (img_width-xmin-1)/ratio)
ymax_diff = random.random() * ratio * min(h, (img_height-ymin-1)/ratio)
new_xmin = round(xmin - xmin_diff)
new_ymin = round(ymin - ymin_diff)
new_xmax = round(xmax + xmax_diff)
new_ymax = round(ymax + ymax_diff)
return new_xmin, new_ymin, new_xmax, new_ymaxSome image of the processed dataset is as follows:
You can also download the data processed directly. And the process script file deal.py is also included.
cd path_to_PaddleClas
Enter the dataset/ directory, download and unzip the dataset.
cd dataset
wget https://paddleclas.bj.bcebos.com/data/PULC/traffic_sign.tar
tar -xf traffic_sign.tar
cd ../The datas under traffic_sign directory:
traffic_sign
├── train
│ ├── 0_62627.jpg
│ ├── 100000_89031.jpg
│ ├── 100001_89031.jpg
...
├── test
│ ├── 100423_2315.jpg
│ ├── 100424_2315.jpg
│ ├── 100425_2315.jpg
...
├── other
│ ├── 100603_3422.jpg
│ ├── 100604_3422.jpg
...
├── label_list_train.txt
├── label_list_test.txt
├── label_list_other.txt
├── label_list_train_for_distillation.txt
├── label_list_train.txt.debug
├── label_list_test.txt.debug
├── label_name_id.txt
├── deal.py
Where train/ and test/ are training set and validation set respectively. The label_list_train.txt and label_list_test.txt are label files of training data and validation data respectively. The file label_list_train.txt.debug and label_list_test.txt.debug are subset of train_list.txt and val_list.txt respectively. other would be used for SKL-UGI knowledge distillation, and its label file is label_list_train_for_distillation.txt.
Note:
- About the contents format of
label_list_train.txtandlabel_list_train.txt, please refer to Description about Classification Dataset in PaddleClas. - About the
label_list_train_for_distillation.txt, please refer to Knowledge Distillation Label.
The details of training config in ppcls/configs/PULC/traffic_sign/PPLCNet_x1_0.yaml. The command about training as follows:
export CUDA_VISIBLE_DEVICES=0,1,2,3
python3 -m paddle.distributed.launch \
--gpus="0,1,2,3" \
tools/train.py \
-c ./ppcls/configs/PULC/traffic_sign/PPLCNet_x1_0.yamlThe best metric of validation data is between 98.0 and 98.2. There would be fluctuations because the data size is small.
After training, you can use the following commands to evaluate the model.
python3 tools/eval.py \
-c ./ppcls/configs/PULC/traffic_sign/PPLCNet_x1_0.yaml \
-o Global.pretrained_model="output/PPLCNet_x1_0/best_model"Among the above command, the argument -o Global.pretrained_model="output/PPLCNet_x1_0/best_model" specify the path of the best model weight file. You can specify other path if needed.
After training, you can use the model that trained to infer. Command is as follow:
python3 tools/infer.py \
-c ./ppcls/configs/PULC/traffic_sign/PPLCNet_x1_0.yaml \
-o Global.pretrained_model=output/DistillationModel/best_modelThe results:
99603_17806.jpg: class id(s): [216, 145, 49, 207, 169], score(s): [1.00, 0.00, 0.00, 0.00, 0.00], label_name(s): ['pm20', 'pm30', 'pm40', 'pl25', 'pm15']
Note:
- Among the above command, argument
-o Global.pretrained_model="output/PPLCNet_x1_0/best_model"specify the path of the best model weight file. You can specify other path if needed. - The default test image is
deploy/images/PULC/traffic_sign/99603_17806.jpg. And you can test other image, only need to specify the argument-o Infer.infer_imgs=path_to_test_image.
SKL-UGI is a simple but effective knowledge distillation algrithem proposed by PaddleClas.
Training the teacher model with hyperparameters specified in ppcls/configs/PULC/traffic_sign/PPLCNet_x1_0.yaml. The command is as follow:
export CUDA_VISIBLE_DEVICES=0,1,2,3
python3 -m paddle.distributed.launch \
--gpus="0,1,2,3" \
tools/train.py \
-c ./ppcls/configs/PULC/traffic_sign/PPLCNet_x1_0.yaml \
-o Arch.name=ResNet101_vdThe best metric of validation data is about 98.59%. The best teacher model weight would be saved in file output/ResNet101_vd/best_model.pdparams.
The training strategy, specified in training config file ppcls/configs/PULC/traffic_sign/PPLCNet_x1_0_distillation.yaml, the teacher model is ResNet101_vd, the student model is PPLCNet_x1_0 and the additional unlabeled training data is validation data of ImageNet1k. The command is as follow:
export CUDA_VISIBLE_DEVICES=0,1,2,3
python3 -m paddle.distributed.launch \
--gpus="0,1,2,3" \
tools/train.py \
-c ./ppcls/configs/PULC/traffic_sign/PPLCNet_x1_0_distillation.yaml \
-o Arch.models.0.Teacher.pretrained=output/ResNet101_vd/best_modelThe best metric is about 98.35%. The best student model weight would be saved in file output/DistillationModel/best_model_student.pdparams.
The hyperparameters used by 3.2 section and 4.1 section are according by Hyperparameters Searching in PaddleClas. If you want to get better results on your own dataset, you can refer to Hyperparameters Searching to get better hyperparameters.
Note: This section is optional. Because the search process will take a long time, you can selectively run according to your specific. If not replace the dataset, you can ignore this section.
Paddle Inference is the original Inference Library of the PaddlePaddle, provides high-performance inference for server deployment. And compared with directly based on the pretrained model, Paddle Inference can use tools to accelerate prediction, so as to achieve better inference performance. Please refer to Paddle Inference for more information.
Paddle Inference need Paddle Inference Model to predict. Two process provided to get Paddle Inference Model. If want to use the provided by PaddleClas, you can download directly, click Downloading Inference Model.
The command about exporting Paddle Inference Model is as follow:
python3 tools/export_model.py \
-c ./ppcls/configs/PULC/traffic_sign/PPLCNet_x1_0.yaml \
-o Global.pretrained_model=output/DistillationModel/best_model_student \
-o Global.save_inference_dir=deploy/models/PPLCNet_x1_0_traffic_sign_inferAfter running above command, the inference model files would be saved in deploy/models/PPLCNet_x1_0_traffic_sign_infer, as shown below:
├── PPLCNet_x1_0_traffic_sign_infer
│ ├── inference.pdiparams
│ ├── inference.pdiparams.info
│ └── inference.pdmodel
Note: The best model is from knowledge distillation training. If knowledge distillation training is not used, the best model would be saved in output/PPLCNet_x1_0/best_model.pdparams.
You can also download directly.
cd deploy/models
# download the inference model and decompression
wget https://paddleclas.bj.bcebos.com/models/PULC/traffic_sign_infer.tar && tar -xf traffic_sign_infer.tar
After decompression, the directory models should be shown below.
├── traffic_sign_infer
│ ├── inference.pdiparams
│ ├── inference.pdiparams.info
│ └── inference.pdmodel
Return the directory deploy:
cd ../
Run the following command to classify traffic sign about the image ./images/PULC/traffic_sign/99603_17806.jpg.
# Use the following command to predict with GPU.
python3.7 python/predict_cls.py -c configs/PULC/traffic_sign/inference_traffic_sign.yaml
# Use the following command to predict with CPU.
python3.7 python/predict_cls.py -c configs/PULC/traffic_sign/inference_traffic_sign.yaml -o Global.use_gpu=FalseThe prediction results:
99603_17806.jpg: class id(s): [216, 145, 49, 207, 169], score(s): [1.00, 0.00, 0.00, 0.00, 0.00], label_name(s): ['pm20', 'pm30', 'pm40', 'pl25', 'pm15']
If you want to predict images in directory, please specify the argument Global.infer_imgs as directory path by -o Global.infer_imgs. The command is as follow.
# Use the following command to predict with GPU. If want to replace with CPU, you can add argument -o Global.use_gpu=False
python3.7 python/predict_cls.py -c configs/PULC/traffic_sign/inference_traffic_sign.yaml -o Global.infer_imgs="./images/PULC/traffic_sign/"All prediction results will be printed, as shown below.
100999_83928.jpg: class id(s): [182, 179, 162, 128, 24], score(s): [0.99, 0.01, 0.00, 0.00, 0.00], label_name(s): ['pl110', 'pl100', 'pl120', 'p26', 'pm10']
99603_17806.jpg: class id(s): [216, 145, 49, 24, 169], score(s): [1.00, 0.00, 0.00, 0.00, 0.00], label_name(s): ['pm20', 'pm30', 'pm40', 'pm10', 'pm15']
About the label_name details, please refer to dataset/traffic_sign/report.pdf.
PaddleClas provides an example about how to deploy with C++. Please refer to Deployment with C++.
Paddle Serving is a flexible, high-performance carrier for machine learning models, and supports different protocol, such as RESTful, gRPC, bRPC and so on, which provides different deployment solutions for a variety of heterogeneous hardware and operating system environments. Please refer Paddle Serving for more information.
PaddleClas provides an example about how to deploy as service by Paddle Serving. Please refer to Paddle Serving Deployment.
Paddle-Lite is an open source deep learning framework that designed to make easy to perform inference on mobile, embeded, and IoT devices. Please refer to Paddle-Lite for more information.
PaddleClas provides an example of how to deploy on mobile by Paddle-Lite. Please refer to Paddle-Lite deployment.
Paddle2ONNX support convert Paddle Inference model to ONNX model. And you can deploy with ONNX model on different inference engine, such as TensorRT, OpenVINO, MNN/TNN, NCNN and so on. About Paddle2ONNX details, please refer to Paddle2ONNX.
PaddleClas provides an example of how to convert Paddle Inference model to ONNX model by paddle2onnx toolkit and predict by ONNX model. You can refer to paddle2onnx for deployment details.