Rotated Object Detection Module Usage Tutorial¶

I. Overview¶

Rotated object detection is a derivative of the object detection module, specifically designed for detecting rotated objects. Rotated bounding boxes (Rotated Bounding Boxes) are commonly used for detecting rectangles with angle information, where the width and height of the rectangle are no longer parallel to the image coordinate axes. Compared to horizontal bounding boxes, rotated bounding boxes generally include less background information. Rotated box detection is often used in scenarios such as remote sensing.

II. Supported Model List¶

Model	Model Download Link	mAP(%)	GPU Inference Time (ms) [Normal Mode / High-Performance Mode]	CPU Inference Time (ms)	Model Storage Size (M)	Introduction
PP-YOLOE-R-L	Inference Model/Training Model	78.14	20.7039	157.942	211.0 M	PP-YOLOE-R is an efficient single-stage Anchor-free rotated box detection model. Based on PP-YOLOE, PP-YOLOE-R introduces a series of useful designs to improve detection accuracy with minimal parameters and computational cost.

Test Environment Description:

Performance Test Environment
Test Dataset: DOTA validation set.
Hardware Configuration:
- GPU: NVIDIA Tesla T4
- CPU: Intel Xeon Gold 6271C @ 2.60GHz
- Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
Inference Mode 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 details, please refer to PaddleX Local Installation Tutorial

After completing the installation of the wheel package, a few lines of code can complete the inference of the rotated object detection module. You can switch models under this module at will, and you can also integrate the model inference of the rotated object detection module into your project. Before running the following code, please download the sample image to your local machine.

from paddlex import create_model
model = create_model("PP-YOLOE-R-L")
output = model.predict("rotated_object_detection_001.png", batch_size=1)
for res in output:
    res.print()
    res.save_to_img("./output/")
    res.save_to_json("./output/res.json")

After running, the result obtained is:

{'res': "{'input_path': 'rotated_object_detection_001.png', 'boxes': [{'cls_id': 4, 'label': 'small-vehicle', 'score': 0.7513620853424072, 'coordinate': [92.72234, 763.36676, 84.7699, 749.9725, 116.207375, 731.8547, 124.15982, 745.2489]}, {'cls_id': 4, 'label': 'small-vehicle', 'score': 0.7284387350082397, 'coordinate': [348.60703, 177.85127, 332.80432, 149.83975, 345.37347, 142.95677, 361.17618, 170.96828]}, {'cls_id': 11, 'label': 'roundabout', 'score': 0.7909174561500549, 'coordinate': [535.02216, 697.095, 201.49803, 608.4738, 292.2446, 276.9634, 625.76874, 365.5845]}]}"}

The meanings of the parameters in the running results are as follows: - input_path: The path of the input image to be predicted. - boxes: Information about each predicted object. - cls_id: Class ID. - label: Class name. - score: Prediction score. - coordinate: Coordinates of the predicted bounding box, in the format [x1, y1, x2, y2, x3, y3, x4, y4].

The visualization image is as follows:

Note: Due to network issues, the parsing of the above URL may not have been successful. If you need the content of this webpage, please check the validity of the URL and try again.

Related methods and parameter explanations are as follows:

create_model instantiates a rotated object detection model (using PP-YOLOE-R_L as an example). The specific explanations are as follows:

Parameter	Parameter Description	Parameter 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
`threshold`	The threshold for filtering low-score objects	`float/None/dict`	None	None
`img_size`	The resolution used by the model for prediction	`int/tuple/None`	None	None

The model_name must be specified. After specifying model_name, the model parameters built into PaddleX will be used by default. If model_dir is specified, the user-defined model will be used.
threshold is the threshold for filtering low-score objects. The default is None, which means using the settings from the lower priority. The priority of parameter settings from high to low is: predict parameter input > create_model initialization > yaml configuration file setting. Currently, two threshold setting methods are supported:
float: Use the same threshold for all classes.
dict: The key is the class ID, and the value is the threshold, allowing different thresholds for different classes.
img_size is the resolution used by the model for prediction. The default is None, which means using the settings from the lower priority. The priority of parameter settings from high to low is: create_model initialization > yaml configuration file setting.
The predict() method of the rotated object detection model is called for inference prediction. The parameters of the predict() method are input, batch_size, and threshold, with specific explanations 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, the elements of the list must be the above types of data, 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
`threshold`	The threshold for filtering low-score objects	`float`/`dict`/`None`	None, indicating the use of settings from the lower priority. The priority of parameter settings from high to low is: `predict parameter input > create_model initialization > yaml configuration file setting` float, such as 0.5, indicating the use of `0.5` as the threshold for all classes during inference dict, such as `{0: 0.5, 1: 0.35}`, indicating the use of 0.5 as the threshold for class 0 and 0.35 as the threshold for class 1 during inference	None

The prediction results are processed, and the prediction result of each sample is of type dict, supporting 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 and make it more readable. This is 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. This is only effective when `format_json` is `True`	`False`
`save_to_json()`	Save the results as a file in JSON format	`save_path`	`str`	The file path for saving. When 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 and make it more readable. This is 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. This is only effective when `format_json` is `True`	`False`
`save_to_img()`	Save the results as a file in image format	`save_path`	`str`	The file path for saving. When it is a directory, the saved file name will be consistent with the input file name	None

In addition, it also supports obtaining the visualization image with results and the prediction results through attributes, as follows:

Attribute	Attribute Description
`json`	Get the prediction results in `json` format
`img`	Get the visualization image in `dict` format

For more usage methods of the single model inference API in PaddleX, please refer to PaddleX Single Model Python Script Usage Instructions.

IV. Secondary Development¶

If you are pursuing higher accuracy with existing models, you can use PaddleX's secondary development capabilities to develop better rotated object detection models. Before using PaddleX to develop rotated object detection models, please ensure that you have installed the model training plugins related to rotated object detection in PaddleX. The installation process can be referred to PaddleX Local Installation Tutorial

4.1 Data Preparation¶

Before model training, you need to prepare the dataset for the corresponding task module. PaddleX provides data verification functionality for each module, only data that passes the verification can be used for model training. Additionally, PaddleX provides a Demo dataset for each module, which you can use to complete subsequent development. If you wish to use a private dataset for subsequent model training, you can refer to PaddleX Object Detection Task Module Data Annotation Tutorial.

4.1.1 Demo Data Download¶

You can refer to the following command to download the Demo dataset to the specified folder:

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

After decompression, the dataset directory structure is as follows:：

- dataset/DOTA-sampled200_crop1024_data
  - annotations
    - instance_train.json
    - instance_val.json
  - images
    - img1.png
    - img2.png
    - img3.png
    ...

4.1.2 Data Verification¶

A single command can complete data verification:

python main.py -c paddlex/configs/modules/rotated_object_detection/PP-YOLOE-R-L.yaml \
    -o Global.mode=check_dataset \
    -o Global.dataset_dir=./dataset/DOTA-sampled200_crop1024_data

After executing the above command, PaddleX will verify the dataset and count the basic information of the dataset. After the command runs successfully, the log will print Check dataset passed !. The verification result file is saved in ./output/check_dataset_result.json, and the related outputs are saved in the./output/check_datasetdirectory under the current directory, including visualized sample images and sample distribution histograms.

👉 Verification Result Details (Click to Expand)

The specific content of the verification result file is:

{
  "done_flag": true,
  "check_pass": true,
  "attributes": {
    "num_classes": 15,
    "train_samples": 1892,
    "train_sample_paths": [
      "check_dataset\/demo_img\/P2610__1.0__0___0.png",
      "check_dataset\/demo_img\/P1137__1.0__0___0.png",
      "check_dataset\/demo_img\/P1122__1.0__5888___1648.png",
      "check_dataset\/demo_img\/P0543__1.0__0___0.png",
      "check_dataset\/demo_img\/P0518__1.0__0___91.png",
      "check_dataset\/demo_img\/P0961__1.0__1648___87.png",
      "check_dataset\/demo_img\/P1732__1.0__0___824.png",
      "check_dataset\/demo_img\/P2766__1.0__4421___0.png",
      "check_dataset\/demo_img\/P2582__1.0__674___725.png",
      "check_dataset\/demo_img\/P1529__1.0__2976___1648.png"
    ],
    "val_samples": 473,
    "val_sample_paths": [
      "check_dataset\/demo_img\/P2342__1.0__890___0.png",
      "check_dataset\/demo_img\/P1386__1.0__2472___1648.png",
      "check_dataset\/demo_img\/P0961__1.0__824___87.png",
      "check_dataset\/demo_img\/P1651__1.0__824___824.png",
      "check_dataset\/demo_img\/P1529__1.0__824___2976.png",
      "check_dataset\/demo_img\/P0961__1.0__4944___87.png",
      "check_dataset\/demo_img\/P0725__1.0__634___0.png",
      "check_dataset\/demo_img\/P1679__1.0__1648___1648.png",
      "check_dataset\/demo_img\/P2726__1.0__824___1578.png",
      "check_dataset\/demo_img\/P0457__1.0__379___0.png",
    ]
  },
  "analysis": {
    "histogram": "check_dataset/histogram.png"
  },
  "dataset_path": "rdet_dota_examples",
  "show_type": "image",
  "dataset_type": "COCODetDataset"
}

In the above verification result, check_pass is true, indicating that the dataset format meets the requirements. The explanations for other indicators are as follows:

attributes.num_classes：The number of categories in this dataset is 15;
attributes.train_samples：The number of training set samples in this dataset is 1892;
attributes.val_samples：The number of validation set samples in this dataset is 473;
attributes.train_sample_paths：The relative path list of visualized training set sample images in this dataset;
attributes.val_sample_paths：The relative path list of visualized validation set sample images in this dataset;

Additionally, the dataset verification also analyzes the distribution of sample quantities for all categories in the dataset and draws 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 or re-split the training/validation ratio of the dataset by modifying the configuration file or adding hyperparameters.

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

（1）Dataset Format Conversion

Rotated object detection does not support dataset format conversion, only standard DOTA COCO data format

（2）Dataset Splitting

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

CheckDataset:
split:
enable: Whether to re-split the dataset, set to True to convert the dataset format, default is False；
train_percent: If re-splitting the dataset, you need to set the percentage of the training set, which is any integer between 0-100, and needs to ensure that the sum with val_percent is 100；
val_percent: If re-splitting the dataset, you need to set the percentage of the validation set, which is any integer between 0-100, and needs to ensure that the sum with train_percent is 100; For example, if you want to re-split the dataset with 90% for the training set and 10% for the 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/rotated_object_detection/PP-YOLOE-R-L.yaml \
    -o Global.mode=check_dataset \
    -o Global.dataset_dir=./dataset/DOTA-sampled200_crop1024_data

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

The above parameters also support setting through adding command line parameters:

python main.py -c paddlex/configs/modules/rotated_object_detection/PP-YOLOE-R-L.yaml \
    -o Global.mode=check_dataset \
    -o Global.dataset_dir=./dataset/DOTA-sampled200_crop1024_data \
    -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 model training, taking the training of the rotated object detection model PP-YOLOE-R-L as an example:

python main.py -c paddlex/configs/modules/rotated_object_detection/PP-YOLOE-R-L.yaml \
    -o Global.mode=train \
    -o Global.dataset_dir=./dataset/DOTA-sampled200_crop1024_data

The following steps are required:

Specify the path of the model's .yaml configuration file (here it is PP-YOLOE-R-L.yaml. When training other models, you need to specify the corresponding configuration file. The correspondence between models and configuration files can be found in PaddleX Model List (CPU/GPU))）
Specify the mode as model training: -o Global.mode=train
Specify the training dataset path: -o Global.dataset_dir Other related parameters can be set by modifying the fields under Global and Train in the .yaml configuration file, or by adding parameters in the command line. For example, specify the first 2 GPU cards for training: -o Global.device=gpu:0,1; set the number of training epochs to 10: -o Train.epochs_iters=10. For more modifiable parameters and detailed explanations, please refer to the configuration file instructions for the corresponding task module PaddleX Common Model Configuration File Parameter Instructions..

👉 More Explanations (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, .pdmodel: Model weight-related files, including network parameters, optimizer, EMA, static graph network parameters, static graph network structure, etc.;

4.3 Model Evaluation¶

After completing model training, you can evaluate the specified model weights 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/rotated_object_detection/PP-YOLOE-R-L.yaml \
    -o Global.mode=evaluate \
    -o Global.dataset_dir=./dataset/DOTA-sampled200_crop1024_data

Similar to model training, the following steps are required:

Specify the .yaml configuration file path for the model (here it is PP-YOLOE-R-L.yaml)
Specify the mode as 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. For details, refer to PaddleX Common Model Configuration File Parameter Description.

👉 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 generated, which records the evaluation results, specifically whether the evaluation task was completed successfully and the model's evaluation metrics, including AP.

4.4 Model Inference and Integration¶

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

4.4.1 Model Inference¶

To perform inference predictions through the command line, 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/rotated_object_detection/PP-YOLOE-R-L.yaml  \
    -o Global.mode=predict \
    -o Predict.model_dir="./output/best_model/inference" \
    -o Predict.input="rotated_object_detection_001.png"

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

Specify the .yaml configuration file path for the model (here it is PP-YOLOE-R-L.yaml)
Specify the mode as 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, refer to PaddleX Common Model Configuration File Parameter Description.

4.4.2 Model Integration¶

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

2.Module Integration

The weights you produce can be directly integrated into the object detection 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.