--- comments: true --- # Image Classification Module Development Tutorial ## I. Overview The image classification module is a crucial component in computer vision systems, responsible for categorizing input images. The performance of this module directly impacts the accuracy and efficiency of the entire computer vision system. Typically, the image classification module receives an image as input and, through deep learning or other machine learning algorithms, classifies it into predefined categories based on its characteristics and content. For instance, in an animal recognition system, the image classification module might need to classify an input image as "cat," "dog," "horse," etc. The classification results from the image classification module are then output for use by other modules or systems. ## II. List of Supported Models > The inference time only includes the model inference time and does not include the time for pre- or post-processing.
ModelModel Download Link Top1 Acc(%) GPU Inference Time (ms)
[Normal Mode / High-Performance Mode]		 CPU Inference Time (ms)
[Normal Mode / High-Performance Mode]		 Model Storage Size (MB)
CLIP_vit_base_patch16_224 Inference Model/Training Model 85.36 12.03 / 2.49 60.86 / 42.69 331
MobileNetV3_small_x1_0 Inference Model/Training Model 68.2 4.23 / 0.78 5.24 / 1.48 10.5
PP-HGNet_small Inference Model/Training Model 81.51 5.87 / 1.68 25.58 / 18.50 86.5
PP-HGNetV2-B0 Inference Model/Training Model 77.77 4.41 / 0.87 10.58 / 1.87 21.4
PP-HGNetV2-B4 Inference Model/Training Model 83.57 7.05 / 1.16 16.23 / 7.55 70.4
PP-HGNetV2-B6 Inference Model/Training Model 86.30 13.86 / 3.28 67.25 / 56.70 268.4
PP-LCNet_x1_0 Inference Model/Training Model 71.32 2.59 / 0.68 3.18 / 1.19 10.5
ResNet50 Inference Model/Training Model 76.5 6.25 / 1.17 15.93 / 9.72 90.8
SwinTransformer_tiny_patch4_window7_224 Inference Model/Training Model 81.10 7.11 / 2.01 62.72 / 47.35 100.1
> ❗ The above list features the 9 core models that the image classification module primarily supports. In total, this module supports 80 models. The complete list of models is as follows:
👉Details of Model List
ModelModel Download Link Top-1 Accuracy (%) GPU Inference Time (ms)
[Normal Mode / High-Performance Mode]		 CPU Inference Time (ms)
[Normal Mode / High-Performance Mode]		 Model Storage Size (MB) Description
CLIP_vit_base_patch16_224 Inference Model/Training Model 85.36 12.03 / 2.49 60.86 / 42.69 331 CLIP is an image classification model based on the correlation between vision and language. It adopts contrastive learning and pre-training methods to achieve unsupervised or weakly supervised image classification, especially suitable for large-scale datasets. By mapping images and texts into the same representation space, the model learns general features, exhibiting good generalization ability and interpretability. With relatively good training errors, it performs well in many downstream tasks.
CLIP_vit_large_patch14_224 Inference Model/Training Model 88.1 49.15 / 9.75 223.16 / 206.49 1040
ConvNeXt_base_224 Inference Model/Training Model 83.84 11.37 / 5.65 143.98 / 52.31 313.9 The ConvNeXt series of models were proposed by Meta in 2022, based on the CNN architecture. This series of models builds upon ResNet, incorporating the advantages of SwinTransformer, including training strategies and network structure optimization ideas, to improve the pure CNN architecture network. It explores the performance limits of convolutional neural networks. The ConvNeXt series of models possesses many advantages of convolutional neural networks, including high inference efficiency and ease of migration to downstream tasks.
ConvNeXt_base_384 Inference Model/Training Model 84.90 29.48 / 11.17 293.76 / 134.27 313.9
ConvNeXt_large_224 Inference Model/Training Model 84.26 22.99 / 12.73 220.79 / 113.24 700.7
ConvNeXt_large_384 Inference Model/Training Model 85.27 58.90 / 24.63 509.48 / 260.27 700.7
ConvNeXt_small Inference Model/Training Model 83.13 7.72 / 4.35 95.92 / 33.34 178.0
ConvNeXt_tiny Inference Model/Training Model 82.03 6.00 / 2.47 63.59 / 18.23 104.1
FasterNet-L Inference Model/Training Model 83.5 11.96 / 2.68 51.93 / 35.33 357.1 FasterNet is a neural network designed to improve runtime speed. Its key improvements are as follows:
1. Re-examined popular operators and found that low FLOPS mainly stem from frequent memory accesses, especially in depthwise convolutions;
2. Proposed Partial Convolution (PConv) to extract image features more efficiently by reducing redundant computations and memory accesses;
3. Launched the FasterNet series of models based on PConv, a new design scheme that achieves significantly higher runtime speeds on various devices without compromising model task performance.
FasterNet-M Inference Model/Training Model 83.0 11.17 / 2.16 38.49 / 21.17 204.6
FasterNet-S Inference Model/Training Model 81.3 7.70 / 1.24 19.51 / 11.22 119.3
FasterNet-T0 Inference Model/Training Model 71.9 4.73 / 0.82 6.40 / 1.96 15.1
FasterNet-T1 Inference Model/Training Model 75.9 4.80 / 0.80 8.14 / 3.13 29.2
FasterNet-T2 Inference Model/Training Model 79.1 6.10 / 0.88 12.71 / 5.35 57.4
MobileNetV1_x0_5 Inference Model/Training Model 63.5 1.98 / 0.51 2.50 / 1.04 4.8 MobileNetV1 is a network released by Google in 2017 for mobile devices or embedded devices. This network decomposes traditional convolution operations into depthwise separable convolutions, which are a combination of Depthwise convolution and Pointwise convolution. Compared to traditional convolutional networks, this combination can significantly reduce the number of parameters and computations. Additionally, this network can be used for image classification and other vision tasks.
MobileNetV1_x0_25 Inference Model/Training Model 51.4 1.99 / 0.45 1.82 / 0.73 1.8
MobileNetV1_x0_75 Inference Model/Training Model 68.8 2.33 / 0.41 3.33 / 1.34 9.3
MobileNetV1_x1_0 Inference Model/Training Model 71.0 2.31 / 0.45 3.91 / 1.89 15.2
MobileNetV2_x0_5 Inference Model/Training Model 65.0 3.58 / 0.62 3.86 / 1.23 7.1 MobileNetV2 is a lightweight network proposed by Google following MobileNetV1. Compared to MobileNetV1, MobileNetV2 introduces Linear bottlenecks and Inverted residual blocks as the basic structure of the network. By stacking these basic modules extensively, the network structure of MobileNetV2 is formed. Finally, it achieves higher classification accuracy with only half the FLOPs of MobileNetV1.
MobileNetV2_x0_25 Inference Model/Training Model 53.2 3.05 / 0.66 3.30 / 0.98 5.5
MobileNetV2_x1_0 Inference Model/Training Model 72.2 3.85 / 0.63 5.50 / 1.87 12.6
MobileNetV2_x1_5 Inference Model/Training Model 74.1 3.93 / 0.73 8.84 / 3.12 25.0
MobileNetV2_x2_0 Inference Model/Training Model 75.2 3.89 / 0.79 10.36 / 4.50 41.2
MobileNetV3_large_x0_5 Inference Model/Training Model 69.2 4.60 / 0.77 5.32 / 1.58 9.6 MobileNetV3 is a NAS-based lightweight network proposed by Google in 2019. To further enhance performance, relu and sigmoid activation functions are replaced with hard_swish and hard_sigmoid activation functions, respectively. Additionally, some improvement strategies specifically designed to reduce network computations are introduced.
MobileNetV3_large_x0_35 Inference Model/Training Model 64.3 4.44 / 0.75 5.20 / 1.50 7.5
MobileNetV3_large_x0_75 Inference Model/Training Model 73.1 5.30 / 0.85 6.02 / 1.93 14.0
MobileNetV3_large_x1_0 Inference Model/Training Model 75.3 5.38 / 0.81 7.16 / 2.19 19.5
MobileNetV3_large_x1_25 Inference Model/Training Model 76.4 5.54 / 0.84 7.06 / 2.84 26.5
MobileNetV3_small_x0_5 Inference Model/Training Model 59.2 3.87 / 0.77 4.90 / 1.32 6.8
MobileNetV3_small_x0_35 Inference Model/Training Model 53.0 3.68 / 0.77 3.94 / 1.27 6.0
MobileNetV3_small_x0_75 Inference Model/Training Model 66.0 3.92 / 0.77 4.68 / 1.39 8.5
MobileNetV3_small_x1_0 Inference Model/Training Model 68.2 4.23 / 0.78 5.24 / 1.48 10.5
MobileNetV3_small_x1_25 Inference Model/Training Model 70.7 4.59 / 0.79 5.36 / 1.63 13.0
MobileNetV4_conv_large Inference Model/Training Model 83.4 9.04 / 2.28 34.34 / 22.01 125.2 MobileNetV4 is an efficient architecture specifically designed for mobile devices. Its core lies in the introduction of the UIB (Universal Inverted Bottleneck) module, a unified and flexible structure that integrates IB (Inverted Bottleneck), ConvNeXt, FFN (Feed Forward Network), and the latest ExtraDW (Extra Depthwise) module. Alongside UIB, Mobile MQA, a customized attention block for mobile accelerators, was also introduced, achieving up to 39% significant acceleration. Furthermore, MobileNetV4 introduces a novel Neural Architecture Search (NAS) scheme to enhance the effectiveness of the search process.
MobileNetV4_conv_medium Inference Model/Training Model 79.9 5.70 / 1.05 13.78 / 5.64 37.6
MobileNetV4_conv_small Inference Model/Training Model 74.6 3.81 / 0.55 5.24 / 1.50 14.7
MobileNetV4_hybrid_large Inference Model/Training Model 83.8 13.43 / 4.28 61.16 / 31.06 145.1
MobileNetV4_hybrid_medium Inference Model/Training Model 80.5 11.82 / 1.30 22.01 / 6.06 42.9
PP-HGNet_base Inference Model/Training Model 85.0 13.43 / 3.81 71.24 / 51.48 249.4 PP-HGNet (High Performance GPU Net) is a high-performance backbone network developed by Baidu PaddlePaddle's vision team, tailored for GPU platforms. This network combines the fundamentals of VOVNet with learnable downsampling layers (LDS Layer), incorporating the advantages of models such as ResNet_vd and PPHGNet. On GPU platforms, this model achieves higher accuracy compared to other SOTA models at the same speed. Specifically, it outperforms ResNet34-0 by 3.8 percentage points and ResNet50-0 by 2.4 percentage points. Under the same SLSD conditions, it ultimately surpasses ResNet50-D by 4.7 percentage points. Additionally, at the same level of accuracy, its inference speed significantly exceeds that of mainstream Vision Transformers.
PP-HGNet_small Inference Model/Training Model 81.51 5.87 / 1.68 25.58 / 18.50 86.5
PP-HGNet_tiny Inference Model/Training Model 79.83 5.84 / 1.38 17.03 / 10.58 52.4
PP-HGNetV2-B0 Inference Model/Training Model 77.77 4.41 / 0.87 10.58 / 1.87 21.4 PP-HGNetV2 (High Performance GPU Network V2) is the next-generation version of Baidu PaddlePaddle's PP-HGNet, featuring further optimizations and improvements upon its predecessor. It pushes the limits of NVIDIA's "Accuracy-Latency Balance," significantly outperforming other models with similar inference speeds in terms of accuracy. It demonstrates strong performance across various label classification and evaluation scenarios.
PP-HGNetV2-B1 Inference Model/Training Model 79.18 4.52 / 0.73 11.98 / 2.28 22.6
PP-HGNetV2-B2 Inference Model/Training Model 81.74 6.67 / 0.96 14.22 / 4.04 39.9
PP-HGNetV2-B3 Inference Model/Training Model 82.98 7.47 / 1.94 17.73 / 5.63 57.9
PP-HGNetV2-B4 Inference Model/Training Model 83.57 7.05 / 1.16 16.23 / 7.55 70.4
PP-HGNetV2-B5 Inference Model/Training Model 84.75 10.38 / 1.95 31.53 / 18.02 140.8
PP-HGNetV2-B6 Inference Model/Training Model 86.30 13.86 / 3.28 67.25 / 56.70 268.4
PP-LCNet_x0_5 Inference Model/Training Model 63.14 2.41 / 0.60 2.54 / 0.90 6.7 PP-LCNet is a lightweight backbone network developed by Baidu PaddlePaddle's vision team. It enhances model performance without increasing inference time, significantly surpassing other lightweight SOTA models.
PP-LCNet_x0_25 Inference Model/Training Model 51.86 2.16 / 0.60 2.73 / 0.77 5.5
PP-LCNet_x0_35 Inference Model/Training Model 58.09 2.18 / 0.60 2.32 / 0.89 5.9
PP-LCNet_x0_75 Inference Model/Training Model 68.18 2.61 / 0.58 3.00 / 1.09 8.4
PP-LCNet_x1_0 Inference Model/Training Model 71.32 2.59 / 0.68 3.18 / 1.19 10.5
PP-LCNet_x1_5 Inference Model/Training Model 73.71 2.60 / 0.68 3.98 / 1.66 16.0
PP-LCNet_x2_0 Inference Model/Training Model 75.18 2.53 / 0.68 5.21 / 2.24 23.2
PP-LCNet_x2_5 Inference Model/Training Model 76.60 2.76 / 0.67 6.78 / 3.20 32.1
PP-LCNetV2_base Inference Model/Training Model 77.05 4.04 / 0.62 6.80 / 2.67 23.7 The PP-LCNetV2 image classification model is the next-generation version of PP-LCNet, self-developed by Baidu PaddlePaddle's vision team. Based on PP-LCNet, it has undergone further optimization and improvements, primarily utilizing re-parameterization strategies to combine depthwise convolutions with varying kernel sizes and optimizing pointwise convolutions, Shortcuts, etc. Without using additional data, the PPLCNetV2_base model achieves over 77% Top-1 Accuracy on the ImageNet dataset for image classification, while maintaining an inference time of less than 4.4 ms on Intel CPU platforms.
PP-LCNetV2_large Inference Model/Training Model 78.51 4.91 / 0.85 10.30 / 5.38 37.3
PP-LCNetV2_small Inference Model/Training Model 73.97 3.07 / 0.60 4.28 / 1.58 14.6
ResNet18_vd Inference Model/Training Model 72.3 2.87 / 0.77 7.91 / 4.64 41.5 The ResNet series of models were introduced in 2015, winning the ILSVRC2015 competition with a top-5 error rate of 3.57%. This network innovatively proposed residual structures, which are stacked to construct the ResNet network. Experiments have shown that using residual blocks can effectively improve convergence speed and accuracy.
ResNet18 Inference Model/Training Model 71.0 2.63 / 0.74 6.30 / 4.16 41.5
ResNet34_vd Inference Model/Training Model 76.0 4.47 / 1.09 14.30 / 8.33 77.3
ResNet34 Inference Model/Training Model 74.6 4.20 / 1.07 12.53 / 7.83 77.3
ResNet50_vd Inference Model/Training Model 79.1 6.66 / 1.23 16.34 / 10.00 90.8
ResNet50 Inference Model/Training Model 76.5 6.25 / 1.17 15.93 / 9.72 90.8
ResNet101_vd Inference Model/Training Model 80.2 11.93 / 2.07 32.47 / 23.62 158.4
ResNet101 Inference Model/Training Model 77.6 13.73 / 2.06 29.69 / 17.72 158.4
ResNet152_vd Inference Model/Training Model 80.6 20.70 / 2.82 43.90 / 27.91 214.3
ResNet152 Inference Model/Training Model 78.3 17.86 / 2.79 46.19 / 26.00 214.2
ResNet200_vd Inference Model/Training Model 80.9 22.55 / 3.54 58.54 / 35.70 266.0
StarNet-S1 Inference Model/Training Model 73.6 6.24 / 0.96 8.78 / 2.44 11.2 StarNet focuses on exploring the untapped potential of "star operations" (i.e., element-wise multiplication) in network design. It reveals that star operations can map inputs to high-dimensional, nonlinear feature spaces, a process akin to kernel tricks but without the need to expand the network size. Consequently, StarNet, a simple yet powerful prototype network, is further proposed, demonstrating exceptional performance and low latency under compact network structures and limited computational resources.
StarNet-S2 Inference Model/Training Model 74.8 4.78 / 0.85 7.24 / 2.48 14.3
StarNet-S3 Inference Model/Training Model 77.0 6.77 / 1.07 9.69 / 3.35 22.2
StarNet-S4 Inference Model/Training Model 79.0 9.01 / 1.48 14.79 / 4.58 28.9
SwinTransformer_base_patch4_window7_224 Inference Model/Training Model 83.37 13.04 / 10.77 133.79 / 118.45 340 SwinTransformer is a novel vision Transformer network that can serve as a general-purpose backbone for computer vision tasks. SwinTransformer consists of a hierarchical Transformer structure represented by shifted windows. Shifted windows restrict self-attention computations to non-overlapping local windows while allowing cross-window connections, thereby enhancing network performance.
SwinTransformer_base_patch4_window12_384 Inference Model/Training Model 84.17 33.99 / 28.42 400.19 / 317.36 311.4
SwinTransformer_large_patch4_window7_224 Inference Model/Training Model 86.19 23.69 / 6.18 198.60 / 177.18 694.8
SwinTransformer_large_patch4_window12_384 Inference Model/Training Model 87.06 68.07 / 14.84 609.07 / 525.72 696.1
SwinTransformer_small_patch4_window7_224 Inference Model/Training Model 83.21 12.17 / 3.51 111.03 / 92.51 175.6
SwinTransformer_tiny_patch4_window7_224 Inference Model/Training Model 81.10 7.11 / 2.01 62.72 / 47.35 100.1
Test Environment Description:
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](../../../installation/installation.en.md). After installing the wheel package, you can complete image classification module inference with just a few lines of code. You can switch between models in this module freely, and you can also integrate the model inference of the image classification module into your project. Before running the following code, please download the [demo image](https://paddle-model-ecology.bj.bcebos.com/paddlex/imgs/demo_image/general_image_classification_001.jpg) to your local machine. ```python from paddlex import create_model model = create_model(model_name="PP-LCNet_x1_0") output = model.predict("general_image_classification_001.jpg", batch_size=1) for res in output: res.print(json_format=False) res.save_to_img("./output/") res.save_to_json("./output/res.json") ``` After running, the result obtained is: ```josn {'res': {'input_path': 'test_imgs/general_image_classification_001.jpg', 'class_ids': [296, 279, 270, 537, 356], 'scores': [0.7915499806404114, 0.0173799991607666, 0.014279999770224094, 0.013009999878704548, 0.01221999991685152], 'label_names': ['ice bear, polar bear, Ursus Maritimus, Thalarctos maritimus', 'Arctic fox, white fox, Alopex lagopus', 'white wolf, Arctic wolf, Canis lupus tundrarum', 'dogsled, dog sled, dog sleigh', 'weasel']}} ``` The meanings of the running results parameters are as follows: - `input_path`: Indicates the path of the input image. - `page_index`: If the input is a PDF file, it indicates which page of the PDF is currently being processed; otherwise, it is `None`. - `class_ids`: Indicates the class IDs of the prediction results. - `scores`: Indicates the confidence scores of the prediction results. - `label_names`: Indicates the class names of the prediction results. The visualization image is as follows: **Note:** Due to network issues, the above URL may not be accessible. If you need to access this link, please check the validity of the URL and try again. If the problem persists, it may be related to the link itself or the network connection. Related methods, parameters, and explanations are as follows: * `create_model` instantiates an image classification model (here, `PP-LCNet_x1_0` is used as an example), and the specific explanations are as follows:
Parameter Parameter Description Parameter Type Options Default Value
model_name Name of the model str None PP-LCNet_x1_0
model_dir Path to store 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. After specifying `model_name`, the default model parameters built into PaddleX are used. If `model_dir` is specified, the user-defined model is used. * The `predict()` method of the image classification model is called for inference prediction. The `predict()` method has parameters `input` and `batch_size`, which are explained as follows:
Parameter Parameter Description Parameter 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 network URL of an image file: Example
  • Local directory, the directory should contain data files to be predicted, such as the local path: /root/data/
  • List，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
* The prediction results are processed, and the prediction result for each sample is of type `dict`. It supports operations such as printing, saving as an image, and saving as a `json` file:
Method Method Description Parameter Parameter Type Parameter Description Default Value
print() Print the results to the terminal format_json bool Whether to format the output content using 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 to escape non-ASCII characters to Unicode. If 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 results as a JSON file save_path str The path to save the file. If it is a directory, the saved file name will be consistent with 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 to escape non-ASCII characters to Unicode. If 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_img() Save the results as an image file save_path str The path to save the file. If it is a directory, the saved file name will be consistent with the input file name None
* Additionally, it supports obtaining the visualization image with results and the prediction results through attributes, as follows:
Attribute Attribute Description
json Get the prediction result in json format
img Get the visualization image in dict format
For more information on using PaddleX's single-model inference APIs, please refer to the [PaddleX Single-Model Python Script Usage Instructions](../../instructions/model_python_API.en.md). ## IV. Custom Development If you are seeking higher accuracy from existing models, you can use PaddleX's custom development capabilities to develop better image classification models. Before using PaddleX to develop image classification models, please ensure that you have installed the relevant model training plugins for image classification in PaddleX. The installation process can be found in the custom development section of the [PaddleX Local Installation Guide](../../../installation/installation.en.md). ### 4.1 Data Preparation Before model training, you need to prepare the dataset for the corresponding task module. PaddleX provides data validation functionality for each module, and only data that passes data 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 your own private dataset for subsequent model training, please refer to the [PaddleX Image Classification Task Module Data Annotation Guide](../../../data_annotations/cv_modules/image_classification.en.md). #### 4.1.1 Demo Data Download You can use the following command to download the demo dataset to a specified folder: ```bash cd /path/to/paddlex wget https://paddle-model-ecology.bj.bcebos.com/paddlex/data/cls_flowers_examples.tar -P ./dataset tar -xf ./dataset/cls_flowers_examples.tar -C ./dataset/ ``` #### 4.1.2 Data Validation One command is all you need to complete data validation: ```bash python main.py -c paddlex/configs/modules/image_classification/PP-LCNet_x1_0.yaml \ -o Global.mode=check_dataset \ -o Global.dataset_dir=./dataset/cls_flowers_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.
👉 Validation Results Details (Click to Expand) 
{
  "done_flag": true,
  "check_pass": true,
  "attributes": {
    "label_file": "dataset/label.txt",
    "num_classes": 102,
    "train_samples": 1020,
    "train_sample_paths": [
      "check_dataset/demo_img/image_01904.jpg",
      "check_dataset/demo_img/image_06940.jpg"
    ],
    "val_samples": 1020,
    "val_sample_paths": [
      "check_dataset/demo_img/image_01937.jpg",
      "check_dataset/demo_img/image_06958.jpg"
    ]
  },
  "analysis": {
    "histogram": "check_dataset/histogram.png"
  },
  "dataset_path": "cls_flowers_examples",
  "show_type": "image",
  "dataset_type": "ClsDataset"
}

The above validation results, with check_pass being True, indicate that the dataset format meets the requirements. Explanations for other indicators are as follows:

Additionally, the dataset validation analyzes the sample number distribution across all classes in the dataset and generates a distribution histogram (histogram.png):

#### 4.1.3 Dataset Format Conversion/Dataset Splitting (Optional) After completing data validation, you can convert the dataset format or re-split the training/validation ratio of the dataset by modifying the configuration file or appending hyperparameters.
👉 Dataset Format Conversion/Dataset Splitting Details (Click to Expand)

(1) Dataset Format Conversion

Image classification does not currently support data conversion.

(2) Dataset Splitting

The parameters for dataset splitting can be set by modifying the fields under CheckDataset in the configuration file. The following are example explanations for some of the parameters in the configuration file:

For example, if you want to re-split the dataset with a 90% training set and a 10% validation set, you need to modify the configuration file as follows:

......
CheckDataset:
  ......
  split:
    enable: True
    train_percent: 90
    val_percent: 10
  ......

Then execute the command:

python main.py -c paddlex/configs/modules/image_classification/PP-LCNet_x1_0.yaml \
    -o Global.mode=check_dataset \
    -o Global.dataset_dir=./dataset/cls_flowers_examples

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

These parameters also support being set through appending command line arguments:

python main.py -c paddlex/configs/modules/image_classification/PP-LCNet_x1_0.yaml \
    -o Global.mode=check_dataset \
    -o Global.dataset_dir=./dataset/cls_flowers_examples \
    -o CheckDataset.split.enable=True \
    -o CheckDataset.split.train_percent=90 \
    -o CheckDataset.split.val_percent=10
### 4.2 Model Training A single command can complete the model training. Taking the training of the image classification model PP-LCNet_x1_0 as an example: ``` python main.py -c paddlex/configs/modules/image_classification/PP-LCNet_x1_0.yaml \ -o Global.mode=train \ -o Global.dataset_dir=./dataset/cls_flowers_examples ``` the following steps are required: * Specify the path of the model's `.yaml` configuration file (here it is `PP-LCNet_x1_0.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)](../../../support_list/models_list.en.md)) * Specify the mode as model training: `-o Global.mode=train` * Specify the path of the training dataset: `-o Global.dataset_dir`. Other related parameters can be set by modifying the fields under `Global` and `Train` in the `.yaml` configuration file, or adjusted by appending parameters in the command line. For example, to specify training on the first 2 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 parameter instructions for the corresponding task module of the model [PaddleX Common Model Configuration File Parameters](../../instructions/config_parameters_common.en.md).
👉 More Details (Click to Expand)
## 4.3 Model Evaluation After completing model training, you can evaluate the specified model weight file on the validation set to verify the model accuracy. Using PaddleX for model evaluation, a single command can complete the model evaluation: ```bash python main.py -c paddlex/configs/modules/image_classification/PP-LCNet_x1_0.yaml \ -o Global.mode=evaluate \ -o Global.dataset_dir=./dataset/cls_flowers_examples ``` Similar to model training, the following steps are required: * Specify the path of the model's `.yaml` configuration file (here it is `PP-LCNet_x1_0.yaml`) * Specify the mode as model evaluation: `-o Global.mode=evaluate` * Specify the path of the validation dataset: `-o Global.dataset_dir` * Other related parameters can be set by modifying the fields under `Global` and `Evaluate` in the `.yaml` configuration. Other related parameters can be set by modifying the fields under `Global` and `Evaluate` in the `.yaml` configuration file. For details, please refer to [PaddleX Common Model Configuration File Parameter Description](../../instructions/config_parameters_common.en.md). * 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)

When evaluating the model, you need to specify the model weight 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 generated, which records the evaluation results. Specifically, it records whether the evaluation task was completed successfully and the model's evaluation metrics, including val.top1, val.top5;

### 4.4 Model Inference and Model Integration After completing model training and evaluation, you can use the Training Model weights for inference predictions 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](https://paddle-model-ecology.bj.bcebos.com/paddlex/imgs/demo_image/general_image_classification_001.jpg) to your local machine. ```bash python main.py -c paddlex/configs/modules/image_classification/PP-LCNet_x1_0.yaml \ -o Global.mode=predict \ -o Predict.model_dir="./output/best_model/inference" \ -o Predict.input="general_image_classification_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-LCNet_x1_0.yaml`) * Specify the mode as model inference prediction: `-o Global.mode=predict` * Specify the model weight 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 fields under `Global` and `Predict` in the `.yaml` configuration file. For details, please refer to [PaddleX Common Model Configuration File Parameter Description](../../instructions/config_parameters_common.en.md). #### 4.4.2 Model Integration The model can be directly integrated into the PaddleX pipelines or directly into your own project. 1.Pipeline Integration The image classification module can be integrated into the [General Image Classification Pipeline](../../../pipeline_usage/tutorials/cv_pipelines/image_classification.en.md) of PaddleX. Simply replace the model path to update the image classification module of the relevant pipeline. In pipeline integration, you can use high-performance inference and serving deployment to deploy your obtained model. 2.Module Integration The weights you produce can be directly integrated into the image classification module. You can refer to the Python example code in Quick Integration and simply replace the model with the path to your Training 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](../../../pipeline_deploy/high_performance_inference.en.md).