comments: true

Image Feature Module Tutorial

I. Overview

The image feature module is one of the important tasks in computer vision, primarily referring to the automatic extraction of useful features from image data using deep learning methods, to facilitate subsequent image retrieval tasks. The performance of this module directly affects the accuracy and efficiency of the subsequent tasks. In practical applications, image features typically output a set of feature vectors, which can effectively represent the content, structure, texture, and other information of the image, and will be passed as input to the subsequent retrieval module for processing.

II. Supported Model List

The inference time only includes the model inference time and does not include the time for pre- or post-processing.

Model	Model Download Link	Recall@1 (%)	GPU Inference Time (ms) [Normal Mode / High-Performance Mode]	CPU Inference Time (ms) [Normal Mode / High-Performance Mode]	Model Storage Size (MB)	Description
PP-ShiTuV2_rec	Inference Model/Training Model	84.2	3.91 / 1.06	6.82 / 2.89	16.3	PP-ShiTuV2 is a general image feature system consisting of three modules: object detection, feature extraction, and vector retrieval. These models are part of the feature extraction module and can be selected based on system requirements.
PP-ShiTuV2_rec_CLIP_vit_base	Inference Model/Training Model	88.69	12.57 / 11.62	67.09 / 67.09	306.6
PP-ShiTuV2_rec_CLIP_vit_large	Inference Model/Training Model	91.03	49.85 / 49.85	229.14 / 229.14	1050

Test Environment Description:

  <li><b>Performance Test Environment</b>
      <ul>
          <li><strong>Test Dataset：</strong>PaddleX Custom Dataset.</li>
          <li><strong>Hardware Configuration:</strong>
              <ul>
                  <li>GPU: NVIDIA Tesla T4</li>
                  <li>CPU: Intel Xeon Gold 6271C @ 2.60GHz</li>
              </ul>
          </li>
          <li><strong>Software Environment:</strong>
              <ul>
                  <li>Ubuntu 20.04 / CUDA 11.8 / cuDNN 8.9 / TensorRT 8.6.1.6</li>
                  <li>paddlepaddle 3.0.0 / paddlex 3.0.3</li>
              </ul>
          </li>
          </li>
      </ul>
  </li>
  <li><b>Inference Mode Description</b></li>

Mode	GPU Configuration	CPU Configuration	Acceleration Technology Combination
Normal Mode	FP32 Precision / No TRT Acceleration	FP32 Precision / 8 Threads	PaddleInference
High-Performance Mode	Optimal combination of pre-selected precision types and acceleration strategies	FP32 Precision / 8 Threads	Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.)

III. Quick Integration

❗ Before quick integration, please install the PaddleX wheel package. For detailed instructions, refer to the PaddleX Local Installation Guide

After installing the wheel package, a few lines of code can complete the inference of the image feature module. You can switch between models under this module freely, and you can also integrate the model inference of the image feature module into your project. Before running the following code, please download the demo image to your local machine.

from paddlex import create_model

model_name = "PP-ShiTuV2_rec"
model = create_model(model_name)
output = model.predict("general_image_recognition_001.jpg", batch_size=1)
for res in output:
    res.print()
    res.save_to_json("./output/res.json")

After running, the result is:

{'res': {'input_path': 'general_image_recognition_001.jpg', 'page_index': None, 'feature': array([ 0.04910231, ..., -0.07126085], dtype=float32)}}

The meanings of the parameters are as follows:

input_path: The path to the input image to be predicted.
feature: The extracted image feature vector, with a dimensionality equal to the model's output feature dimension, which is 512 in this case.

Descriptions of related methods, parameters, etc., are as follows:

The create_model method instantiates an image feature model (using PP-ShiTuV2_rec as an example). Specific descriptions are as follows:

Parameter	Description	Type	Options	Default Value
`model_name`	The name of the model	`str`	None	None
`model_dir`	The storage path of the model	`str`	None	None
`device`	The device used for model inference	`str`	It supports specifying specific GPU card numbers, such as "gpu:0", other hardware card numbers, such as "npu:0", or CPU, such as "cpu".	`gpu:0`
`use_hpip`	Whether to enable the high-performance inference plugin	`bool`	None	`False`
`hpi_config`	High-performance inference configuration	`dict` \| `None`	None	`None`

The model_name must be specified. When model_name is specified, PaddleX's built-in model parameters are used by default. If model_dir is specified, the user-defined model is used.
The predict() method of the image feature model is called for inference. The parameters of the predict() method are input and batch_size, with specific descriptions as follows:

Parameter	Description	Type	Options	Default Value
`input`	Data to be predicted, supporting multiple input types	`Python Var`/`str`/`list`	Python Variable, such as image data represented by `numpy.ndarray` File Path, such as the local path of an image file: `/root/data/img.jpg` URL Link, such as the web URL of an image file: Example Local Directory, the directory should contain the data files to be predicted, such as the local path: `/root/data/` List, the elements of the list should be of the above-mentioned data types, such as `[numpy.ndarray, numpy.ndarray]`, `[\"/root/data/img1.jpg\", \"/root/data/img2.jpg\"]`, `[\"/root/data1\", \"/root/data2\"]`	None
`batch_size`	Batch size	`int`	Any integer	1

Process the prediction results. Each sample's prediction result is of type dict, and supports operations such as printing, saving as an image, and saving as a json file:

Method	Description	Parameter	Type	Explanation	Default Value
`print()`	Print the result to the terminal	`format_json`	`bool`	Whether to format the output content with `JSON` indentation	`True`
		`indent`	`int`	Specify the indentation level to beautify the output `JSON` data, making it more readable. Only effective when `format_json` is `True`	4
		`ensure_ascii`	`bool`	Control whether non-`ASCII` characters are escaped to `Unicode`. When set to `True`, all non-`ASCII` characters will be escaped; `False` retains the original characters. Only effective when `format_json` is `True`	`False`
`save_to_json()`	Save the result as a `json` file	`save_path`	`str`	The file path for saving. When a directory is provided, the saved file name matches the input file name	None
		`indent`	`int`	Specify the indentation level to beautify the output `JSON` data, making it more readable. Only effective when `format_json` is `True`	4
		`ensure_ascii`	`bool`	Control whether non-`ASCII` characters are escaped to `Unicode`. When set to `True`, all non-`ASCII` characters will be escaped; `False` retains the original characters. Only effective when `format_json` is `True`	`False`

Additionally, the prediction results can be accessed via properties, as follows:

Property	Description
`json`	Get the prediction result in `json` format

For more information on using PaddleX's single-model inference APIs, refer to the PaddleX Single Model Python Script Usage Instructions.

IV. Custom Development

If you seek higher accuracy from existing models, you can leverage PaddleX's custom development capabilities to develop better image feature models. Before developing image feature models with PaddleX, ensure you have installed the classification-related model training plugins for PaddleX. The installation process can be found in the PaddleX Local Installation Guide

4.1 Data Preparation

Before model training, you need to prepare the corresponding dataset for the task module. PaddleX provides data validation functionality for each module, and only data that passes validation can be used for model training. Additionally, PaddleX provides demo datasets for each module, which you can use to complete subsequent development. If you wish to use private datasets for model training, refer to PaddleX Image Feature Task Module Data Annotation Tutorial.

4.1.1 Demo Data Download

You can use the following commands to download the demo dataset to a specified folder:

wget https://paddle-model-ecology.bj.bcebos.com/paddlex/data/Inshop_examples.tar -P ./dataset
tar -xf ./dataset/Inshop_examples.tar -C ./dataset/

4.1.2 Data Validation

A single command can complete data validation:

python main.py -c paddlex/configs/modules/image_feature/PP-ShiTuV2_rec.yaml \
    -o Global.mode=check_dataset \
    -o Global.dataset_dir=./dataset/Inshop_examples

After executing the above command, PaddleX will validate the dataset and summarize its basic information. If the command runs successfully, it will print Check dataset passed ! in the log. The validation results file is saved in ./output/check_dataset_result.json, and related outputs are saved in the ./output/check_dataset directory in the current directory, including visual examples of sample images and sample distribution histograms.

👉 Details of Validation Results (Click to Expand)

The specific content of the validation result file is:


  "done_flag": true,
  "check_pass": true,
  "attributes": {
    "train_samples": 1000,
    "train_sample_paths": [
      "check_dataset/demo_img/05_1_front.jpg",
      "check_dataset/demo_img/02_1_front.jpg",
      "check_dataset/demo_img/02_3_back.jpg",
      "check_dataset/demo_img/04_3_back.jpg",
      "check_dataset/demo_img/04_2_side.jpg",
      "check_dataset/demo_img/12_1_front.jpg",
      "check_dataset/demo_img/07_2_side.jpg",
      "check_dataset/demo_img/04_7_additional.jpg",
      "check_dataset/demo_img/04_4_full.jpg",
      "check_dataset/demo_img/01_1_front.jpg"
    ],
    "gallery_samples": 110,
    "gallery_sample_paths": [
      "check_dataset/demo_img/06_2_side.jpg",
      "check_dataset/demo_img/01_4_full.jpg",
      "check_dataset/demo_img/04_7_additional.jpg",
      "check_dataset/demo_img/02_1_front.jpg",
      "check_dataset/demo_img/02_3_back.jpg",
      "check_dataset/demo_img/02_3_back.jpg",
      "check_dataset/demo_img/02_4_full.jpg",
      "check_dataset/demo_img/03_4_full.jpg",
      "check_dataset/demo_img/02_2_side.jpg",
      "check_dataset/demo_img/03_2_side.jpg"
    ],
    "query_samples": 125,
    "query_sample_paths": [
      "check_dataset/demo_img/08_7_additional.jpg",
      "check_dataset/demo_img/01_7_additional.jpg",
      "check_dataset/demo_img/02_4_full.jpg",
      "check_dataset/demo_img/04_4_full.jpg",
      "check_dataset/demo_img/09_7_additional.jpg",
      "check_dataset/demo_img/04_3_back.jpg",
      "check_dataset/demo_img/02_1_front.jpg",
      "check_dataset/demo_img/06_2_side.jpg",
      "check_dataset/demo_img/02_7_additional.jpg",
      "check_dataset/demo_img/02_2_side.jpg"
    ]
  },
  "analysis": {
    "histogram": "check_dataset/histogram.png"
  },
  "dataset_path": "./dataset/Inshop_examples",
  "show_type": "image",
  "dataset_type": "ShiTuRecDataset"
}

In the above validation results, check_pass being True indicates that the dataset format meets the requirements. Explanations for other indicators are as follows: * attributes.train_samples: The number of training samples in this dataset is 1000; * attributes.gallery_samples: The number of gallery (or reference) samples in this dataset is 110; * attributes.query_samples: The number of query samples in this dataset is 125; * attributes.train_sample_paths: A list of relative paths to the visual images of training samples in this dataset; * attributes.gallery_sample_paths: A list of relative paths to the visual images of gallery (or reference) samples in this dataset; * attributes.query_sample_paths: A list of relative paths to the visual images of query samples in this dataset;

Additionally, the dataset verification also analyzes the number of images and image categories within the dataset, and generates a distribution histogram (histogram.png):

4.1.3 Dataset Format Conversion / Dataset Splitting (Optional)

After completing the data verification, you can convert the dataset format and re-split the training/validation ratio by modifying the configuration file or appending hyperparameters.

👉 Details of Format Conversion / Dataset Splitting (Click to Expand)

(1) Dataset Format Conversion

The image feature task supports converting LabelMe format datasets to ShiTuRecDataset format. The parameters for dataset format conversion can be set by modifying the fields under CheckDataset in the configuration file. Some example parameter descriptions in the configuration file are as follows:

CheckDataset:
convert:
enable: Whether to perform dataset format conversion. The image feature task supports converting LabelMe format datasets to ShiTuRecDataset format, default is False;
src_dataset_type: If dataset format conversion is performed, the source dataset format needs to be set, default is null, optional value is LabelMe;

For example, if you want to convert a LabelMe format dataset to ShiTuRecDataset format, you need to modify the configuration file as follows:

cd /path/to/paddlex
wget https://paddle-model-ecology.bj.bcebos.com/paddlex/data/image_classification_labelme_examples.tar -P ./dataset
tar -xf ./dataset/image_classification_labelme_examples.tar -C ./dataset/

......
CheckDataset:
  ......
  convert:
    enable: True
    src_dataset_type: LabelMe
  ......

Then execute the command:

python main.py -c paddlex/configs/modules/image_feature/PP-ShiTuV2_rec.yaml \
    -o Global.mode=check_dataset \
    -o Global.dataset_dir=./dataset/image_classification_labelme_examples

After the data conversion is executed, the original annotation files will be renamed to xxx.bak in the original path.

The above parameters also support being set by appending command line arguments:

python main.py -c paddlex/configs/modules/image_feature/PP-ShiTuV2_rec.yaml \
    -o Global.mode=check_dataset \
    -o Global.dataset_dir=./dataset/image_classification_labelme_examples \
    -o CheckDataset.convert.enable=True \
    -o CheckDataset.convert.src_dataset_type=LabelMe

(2) Dataset Splitting

The parameters for dataset splitting can be set by modifying the fields under CheckDataset in the configuration file. Some example parameter descriptions in the configuration file are as follows:

CheckDataset:
split:
enable: Whether to re-split the dataset. When True, the dataset will be re-split, default is False;
train_percent: If the dataset is re-split, the percentage of the training set needs to be set, the type is any integer between 0-100, and it needs to ensure that the sum of gallery_percent and query_percent values is 100;

For example, if you want to re-split the dataset with 70% training set, 20% gallery set, and 10% query set, you need to modify the configuration file as follows:

......
CheckDataset:
  ......
  split:
    enable: True
    train_percent: 70
    gallery_percent: 20
    query_percent: 10
  ......

Then execute the command:

python main.py -c paddlex/configs/modules/image_feature/PP-ShiTuV2_rec.yaml \
    -o Global.mode=check_dataset \
    -o Global.dataset_dir=./dataset/Inshop_examples

After the data splitting is executed, the original annotation files will be renamed to xxx.bak in the original path.

The above parameters also support being set by appending command line arguments:

python main.py -c paddlex/configs/modules/image_feature/PP-ShiTuV2_rec.yaml  \
    -o Global.mode=check_dataset \
    -o Global.dataset_dir=./dataset/Inshop_examples \
    -o CheckDataset.split.enable=True \
    -o CheckDataset.split.train_percent=70 \
    -o CheckDataset.split.gallery_percent=20 \
    -o CheckDataset.split.query_percent=10

❗Note: Due to the specificity of image feature model evaluation, data partitioning is meaningful only when the train, query, and gallery sets belong to the same category system. During the evaluation of recognition models, it is imperative that the gallery and query sets belong to the same category system, which may or may not be the same as the train set. If the gallery and query sets do not belong to the same category system as the train set, the evaluation after data partitioning becomes meaningless. It is recommended to proceed with caution.

4.2 Model Training

Model training can be completed with a single command, taking the training of the image feature model PP-ShiTuV2_rec as an example:

python main.py -c paddlex/configs/modules/image_feature/PP-ShiTuV2_rec.yaml \
    -o Global.mode=train \
    -o Global.dataset_dir=./dataset/Inshop_examples

The following steps are required:

Specify the .yaml configuration file path for the model (here it is PP-ShiTuV2_rec.yaml，When training other models, you need to specify the corresponding configuration files. The relationship between the model and configuration files can be found in the PaddleX Model List (CPU/GPU))
Set the mode to model training: -o Global.mode=train
Specify the path to the training dataset: -o Global.dataset_dir.
Other related parameters can be set by modifying the Global and Train fields in the .yaml configuration file, or adjusted by appending parameters in the command line. For example, to specify training on the first two GPUs: -o Global.device=gpu:0,1; to set the number of training epochs to 10: -o Train.epochs_iters=10. For more modifiable parameters and their detailed explanations, refer to the configuration file instructions for the corresponding task module of the model PaddleX Common Configuration File Parameters.
New Feature: Paddle 3.0 support CINN (Compiler Infrastructure for Neural Networks) to accelerate training speed when using GPU device. Please specify -o Train.dy2st=True to enable it.

👉 More Details (Click to Expand)

During model training, PaddleX automatically saves the model weight files, with the default being output. If you need to specify a save path, you can set it through the -o Global.output field in the configuration file.
PaddleX shields you from the concepts of dynamic graph weights and static graph weights. During model training, both dynamic and static graph weights are produced, and static graph weights are selected by default for model inference.
After completing the model training, all outputs are saved in the specified output directory (default is ./output/), typically including:
train_result.json: Training result record file, recording whether the training task was completed normally, as well as the output weight metrics, related file paths, etc.;
train.log: Training log file, recording changes in model metrics and loss during training;
config.yaml: Training configuration file, recording the hyperparameter configuration for this training session;
.pdparams, .pdema, .pdopt.pdstate, .pdiparams, .json: Model weight-related files, including network parameters, optimizer, EMA, static graph network parameters, static graph network structure, etc.;
Notice: Since Paddle 3.0.0, the format of storing static graph network structure has changed to json(the current.json file) from protobuf(the former.pdmodel file) to be compatible with PIR and more flexible and scalable.

4.3 Model Evaluation

After completing model training, you can evaluate the specified model weight file on the validation set to verify the model's accuracy. Using PaddleX for model evaluation can be done with a single command:

python main.py -c paddlex/configs/modules/image_feature/PP-ShiTuV2_rec.yaml \
    -o Global.mode=evaluate \
    -o Global.dataset_dir=./dataset/Inshop_examples

Similar to model training, the following steps are required:

Specify the .yaml configuration file path for the model (here it is PP-ShiTuV2_rec.yaml)
Set the mode to model evaluation: -o Global.mode=evaluate
Specify the path to the validation dataset: -o Global.dataset_dir. Other related parameters can be set by modifying the Global and Evaluate fields in the .yaml configuration file, detailed instructions can be found in PaddleX Common Configuration File Parameters.

👉 More Details (Click to Expand)

When evaluating the model, you need to specify the model weights file path. Each configuration file has a default weight save path built-in. If you need to change it, simply set it by appending a command line parameter, such as -o Evaluate.weight_path=./output/best_model/best_model.pdparams.

After completing the model evaluation, an evaluate_result.json file will be produced, which records the evaluation results, specifically, whether the evaluation task was completed successfully and the model's evaluation metrics, including recall1、recall5、mAP；

4.4 Model Inference and Integration

After completing model training and evaluation, you can use the trained model weights for inference prediction or Python integration.

4.4.1 Model Inference

To perform inference prediction through the command line, simply use the following command. Before running the following code, please download the demo image to your local machine.

python main.py -c paddlex/configs/modules/image_feature/PP-ShiTuV2_rec.yaml  \
    -o Global.mode=predict \
    -o Predict.model_dir="./output/best_model/inference" \
    -o Predict.input="general_image_recognition_001.jpg"

Similar to model training and evaluation, the following steps are required:

Specify the .yaml configuration file path for the model (here it is PP-ShiTuV2_rec.yaml)
Set the mode to model inference prediction: -o Global.mode=predict
Specify the model weights path: -o Predict.model_dir="./output/best_model/inference"
Specify the input data path: -o Predict.input="...". Other related parameters can be set by modifying the Global and Predict fields in the .yaml configuration file. For details, please refer to PaddleX Common Model Configuration File Parameter Description.

❗ Note: The inference result of the recognition model is a set of vectors, which requires a retrieval module to complete image feature.

4.4.2 Model Integration

The model can be directly integrated into the PaddleX pipeline or directly into your own project.

1.Pipeline Integration

The image feature module can be integrated into the General Image Recognition Pipeline (coming soon) of PaddleX. Simply replace the model path to update the image feature module of the relevant pipeline. In pipeline integration, you can use serving deployment to deploy your trained model.

2.Module Integration

The weights you produce can be directly integrated into the image feature module. Refer to the Python example code in Quick Integration, and simply replace the model with the path to your trained model.

You can also use the PaddleX high-performance inference plugin to optimize the inference process of your model and further improve efficiency. For detailed procedures, please refer to the PaddleX High-Performance Inference Guide.

image_feature.en.md 28 KB History Raw