Sfoglia il codice sorgente

add 3d-related docs

Jonathans575 9 mesi fa
parent
commit
c04cb5f2ab

+ 227 - 0
docs/module_usage/instructions/config_parameters_3d.en.md

@@ -0,0 +1,227 @@
+---
+comments: true
+---
+
+# PaddleX 3d Task Model Configuration File Parameters Explanation
+
+# Global
+
+<table>
+<thead>
+<tr>
+<th>Parameter Name</th>
+<th>Data Type</th>
+<th>Description</th>
+<th>Default Value</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+<td>model</td>
+<td>str</td>
+<td>Specifies the model name</td>
+<td>Model name specified in the YAML file</td>
+</tr>
+<tr>
+<td>mode</td>
+<td>str</td>
+<td>Specifies the mode (check_dataset/train/evaluate/export/predict)</td>
+<td>check_dataset</td>
+</tr>
+<tr>
+<td>dataset_dir</td>
+<td>str</td>
+<td>Path to the dataset</td>
+<td>Dataset path specified in the YAML file</td>
+</tr>
+<tr>
+<td>device</td>
+<td>str</td>
+<td>Specifies the device to use</td>
+<td>Device ID specified in the YAML file</td>
+</tr>
+<tr>
+<td>output</td>
+<td>str</td>
+<td>Output path</td>
+<td>"output"</td>
+</tr>
+<tr>
+<td>load_cam_from</td>
+<td>str</td>
+<td>Pre-trained parameter path of cam branch</td>
+<td>The pre-trained parameter path of the cam branch specified in the YAML file</td>
+</tr>
+<tr>
+<td>load_lidar_from</td>
+<td>str</td>
+<td>Pre-trained parameter path of lidar branch</td>
+<td>The pre-trained parameter path of the lidar branch specified in the YAML file</td>
+</tr>
+<tr>
+<td>datart_prefix</td>
+<td>bool</td>
+<td>Whether to add the root path to the dataset_dir</td>
+<td>True</td>
+</tr>
+<tr>
+<td>version</td>
+<td>str</td>
+<td>Dataset version</td>
+<td>"mini"</td>
+</tr>
+</tbody>
+</table>
+
+# CheckDataset
+
+<table>
+<thead>
+<tr>
+<th>Parameter Name</th>
+<th>Data Type</th>
+<th>Description</th>
+<th>Default Value</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+<td>convert.enable</td>
+<td>bool</td>
+<td>Whether to convert the dataset format</td>
+<td>False (Currently not supported)</td>
+</tr>
+<tr>
+<td>split.enable</td>
+<td>bool</td>
+<td>Whether to re-split the dataset</td>
+<td>False (Currently not supported)</td>
+</tr>
+
+</tbody>
+</table>
+
+# Train
+### Common Parameters for 3d Tasks
+<table>
+<thead>
+<tr>
+<th>Parameter Name</th>
+<th>Data Type</th>
+<th>Description</th>
+<th>Default Value</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+<td>epochs_iters</td>
+<td>int</td>
+<td>The number of times the model repeats learning the training data</td>
+<td>Number of iterations specified in the YAML file</td>
+</tr>
+<tr>
+<td>batch_size</td>
+<td>int</td>
+<td>Batch size</td>
+<td>Batch size specified in the YAML file</td>
+</tr>
+<tr>
+<td>learning_rate</td>
+<td>float</td>
+<td>Initial learning rate</td>
+<td>Initial learning rate specified in the YAML file</td>
+</tr>
+<tr>
+<td>warmup_steps</td>
+<td>str</td>
+<td>The number of steps performed in the warm-up training</td>
+<td>The number of warm-up steps specified in the YAML file</td>
+</tr>
+</tbody>
+</table>
+
+
+# Evaluate
+<table>
+<thead>
+<tr>
+<th>Parameter Name</th>
+<th>Data Type</th>
+<th>Description</th>
+<th>Default Value</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+<td>batch_size</td>
+<td>int</td>
+<td>Batch size</td>
+<td>Batch size specified in the YAML file
+</td>
+</tr>
+<tr>
+<td>weight_path</td>
+<td>str</td>
+<td>Evaluation model path</td>
+<td>Default local path from training output, when specified as None, indicates using official weights</td>
+</tr>
+</tbody>
+</table>
+
+# Export
+<table>
+<thead>
+<tr>
+<th>Parameter Name</th>
+<th>Data Type</th>
+<th>Description</th>
+<th>Default Value</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+<td>weight_path</td>
+<td>str</td>
+<td>Dynamic graph weight path for exporting the model</td>
+<td>Default local path from training output, when specified as None, indicates using official weights</td>
+</tr>
+</tbody>
+</table>
+
+# Predict
+<table>
+<thead>
+<tr>
+<th>Parameter Name</th>
+<th>Data Type</th>
+<th>Description</th>
+<th>Default Value</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+<td>batch_size</td>
+<td>int</td>
+<td>Prediction batch size</td>
+<td>The prediction batch size specified in the YAML file</td>
+</tr>
+<tr>
+<td>model_dir</td>
+<td>str</td>
+<td>Path to the prediction model</td>
+<td>The default local inference model path produced by training. When specified as None, it indicates the use of official weights</td>
+</tr>
+<tr>
+<td>input</td>
+<td>str</td>
+<td>Path to the prediction input</td>
+<td>The prediction input path specified in the YAML file</td>
+</tr>
+<tr>
+<td>kernel_option.run_mode</td>
+<td>str</td>
+<td>Inference engine setting, such as: "paddle"</td>
+<td>paddle</td>
+</tr>
+</tbody>
+</table>

+ 223 - 0
docs/module_usage/instructions/config_parameters_3d.md

@@ -0,0 +1,223 @@
+---
+comments: true
+---
+
+# PaddleX 3d任务模型配置文件参数说明
+
+# Global
+<table>
+<thead>
+<tr>
+<th>参数名</th>
+<th>数据类型</th>
+<th>描述</th>
+<th>默认值</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+<td>model</td>
+<td>str</td>
+<td>指定模型名称</td>
+<td>yaml文件中指定的模型名称</td>
+</tr>
+<tr>
+<td>mode</td>
+<td>str</td>
+<td>指定模式(check_dataset/train/evaluate/export/predict)</td>
+<td>check_dataset</td>
+</tr>
+<tr>
+<td>dataset_dir</td>
+<td>str</td>
+<td>数据集路径</td>
+<td>yaml文件中指定的数据集路径</td>
+</tr>
+<tr>
+<td>device</td>
+<td>str</td>
+<td>指定使用的设备</td>
+<td>yaml文件中指定的设备id</td>
+</tr>
+<tr>
+<td>output</td>
+<td>str</td>
+<td>输出路径</td>
+<td>"output"</td>
+</tr>
+<tr>
+<td>load_cam_from</td>
+<td>str</td>
+<td>cam分支的预训练参数路径</td>
+<td>yaml文件中指定的cam分支的预训练参数路径</td>
+</tr>
+<tr>
+<td>load_lidar_from</td>
+<td>str</td>
+<td>lidar分支的预训练参数路径</td>
+<td>yaml文件中指定的lidar分支的预训练参数路径</td>
+</tr>
+<tr>
+<td>datart_prefix</td>
+<td>bool</td>
+<td>数据集路径是否需要加上根路径</td>
+<td>True</td>
+</tr>
+<tr>
+<td>version</td>
+<td>str</td>
+<td>数据集版本号</td>
+<td>"mini"</td>
+</tr>
+</tbody>
+</table>
+
+# CheckDataset
+<table>
+<thead>
+<tr>
+<th>参数名</th>
+<th>数据类型</th>
+<th>描述</th>
+<th>默认值</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+<td>convert.enable</td>
+<td>bool</td>
+<td>是否进行数据集格式转换</td>
+<td>False (当前不支持)</td>
+</tr>
+<tr>
+<td>split.enable</td>
+<td>bool</td>
+<td>是否重新划分数据集</td>
+<td>False (当前不支持)</td>
+</tr>
+</tbody>
+</table>
+
+# Train
+### 3d任务公共参数
+<table>
+<thead>
+<tr>
+<th>参数名</th>
+<th>数据类型</th>
+<th>描述</th>
+<th>默认值</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+<td>epochs_iters</td>
+<td>int</td>
+<td>模型对训练数据的重复学习次数</td>
+<td>yaml文件中指定的重复学习次数</td>
+</tr>
+<tr>
+<td>batch_size</td>
+<td>int</td>
+<td>批大小</td>
+<td>yaml文件中指定的批大小</td>
+</tr>
+<tr>
+<td>learning_rate</td>
+<td>float</td>
+<td>初始学习率</td>
+<td>yaml文件中指定的初始学习率</td>
+</tr>
+<tr>
+<td>warmup_steps</td>
+<td>str</td>
+<td>训练模型的开始阶段,进行热身训练的步数</td>
+<td>yaml文件中指定的热身训练的步数</td>
+</tr>
+</tbody>
+</table>
+
+# Evaluate
+<table>
+<thead>
+<tr>
+<th>参数名</th>
+<th>数据类型</th>
+<th>描述</th>
+<th>默认值</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+<td>batch_size</td>
+<td>int</td>
+<td>批大小</td>
+<td>yaml文件中指定的批大小
+</td>
+</tr>
+<tr>
+<td>weight_path</td>
+<td>str</td>
+<td>评估模型路径</td>
+<td>默认训练产出的本地路径,当指定为None时,表示使用官方权重</td>
+</tr>
+</tbody>
+</table>
+
+# Export
+<table>
+<thead>
+<tr>
+<th>参数名</th>
+<th>数据类型</th>
+<th>描述</th>
+<th>默认值</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+<td>weight_path</td>
+<td>str</td>
+<td>导出模型的动态图权重路径</td>
+<td>默认训练产出的本地路径,当指定为None时,表示使用官方权重</td>
+</tr>
+</tbody>
+</table>
+
+# Predict
+<table>
+<thead>
+<tr>
+<th>参数名</th>
+<th>数据类型</th>
+<th>描述</th>
+<th>默认值</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+<td>batch_size</td>
+<td>int</td>
+<td>预测批大小</td>
+<td>yaml文件中指定的预测批大小</td>
+</tr>
+<tr>
+<td>model_dir</td>
+<td>str</td>
+<td>预测模型路径</td>
+<td>默认训练产出的本地推理模型路径,当指定为None时,表示使用官方权重</td>
+</tr>
+<tr>
+<td>input</td>
+<td>str</td>
+<td>预测输入路径</td>
+<td>yaml文件中指定的预测输入路径</td>
+</tr>
+<tr>
+<td>kernel_option.run_mode</td>
+<td>str</td>
+<td>推理引擎设置,如: "paddle"</td>
+<td>paddle</td>
+</tr>
+</tbody>
+</table>