文档首页> 自动驾驶云服务 Octopus> 用户指南> 训练服务> 模型评测> 模型数据集支持> 目标检测2D
更新时间：2024-05-11 GMT+08:00
查看PDF
分享

目标检测2D

Octopus

  • 目录

    标注文件目录结构

    +--- 1599625710056
|   +--- 1599625710056.jpg
|   +--- 1599625710056.json
+--- 1599625740054
|   +--- 1599625740054.jpg
|   +--- 1599625740054.json

    推理文件目录结构

    +--- 1599625710056
|   +--- 1599625710056.json
+--- 1599625740054
|   +--- 1599625740054.json
  • 示例标注文件

    2D目标检测-Octopus标注.json

    {
  "labels": [
    {
      "label_meta_name": "Car",
      "bndbox": {
        "ymin": 503,
        "xmax": 980,
        "ymax": 524,
        "xmin": 956
      }
    },
    {
      "label_meta_name": "Car",
      "bndbox": {
        "ymin": 508,
        "xmax": 931,
        "ymax": 531,
        "xmin": 904
      }
    }
  ]
}
  • 示例推理文件

    2D目标检测-Octopus推理.json

    {
  "labels": [
    {
      "label_meta_name": "Car",
      "score": 0.89,
      "bndbox": {
        "ymin": 500,
        "xmax": 970,
        "ymax": 520,
        "xmin": 950
      }
    },
    {
      "label_meta_name": "Car",
      "score": 0.82,
      "bndbox": {
        "ymin": 508,
        "xmax": 930,
        "ymax": 520,
        "xmin": 900
      }
    }
  ]
}

KITTI

  • 目录

    标注文件目录结构

    +--- image_2
|   +--- 000000.png
|   +--- 000001.png
|   +--- 000002.png
+--- label_2
|   +--- 000000.txt
|   +--- 000001.txt
|   +--- 000002.txt

    推理文件目录结构

    +--- 000000.txt
+--- 000001.txt
+--- 000002.txt
  • 示例标注/推理文件
    2D目标检测-KITTI.txt 
    Car 0.00 0 -1.57 956 503 980 524 2.61 2.77 5.92 3.39 54.15 -0.01 -1.55 0.
Car 0.00 0 -1.57 904 508 931 531 2.36 1.41 11.72 -3.22 40.29 1.14 -1.57 0.
Car 0.00 0 -1.57 1118 514 1149 538 1.138 2.723 10.583 -0.636 40.702 0.747 -1.57 0.
Car 0.00 0 -1.57 1018 497 1038 520 1.32 1.14 3.78 2.43 37.73 -1.08 -1.57 0.
Car 0.00 0 -1.57 1048 508 1078 536 1.61 0.75 5.70 -3.11 28.20 0.63 -1.59 0.
Car 0.00 0 -1.57 1175 517 1235 557 1.75 1.96 3.75 -7.57 12.66 -0.67 -1.54 0.
Car 0.00 0 -1.57 756 517 802 552 2.61 2.77 5.92 3.39 54.15 -0.01 -1.55 0.
Car 0.00 0 -1.57 304 533 446 612 2.36 1.41 11.72 -3.22 40.29 1.14 -1.57 0.
Car 0.00 0 -1.57 0 491 310 766 1.138 2.723 10.583 -0.636 40.702 0.747 -1.57 0.
Truck 0.00 0 -1.57 510 442 697 601 1.32 1.14 3.78 2.43 37.73 -1.08 -1.57 0.
Car 0.00 0 -1.57 879 516 917 546 1.61 0.75 5.70 -3.11 28.20 0.63 -1.59 0.
Car 0.00 0 -1.57 986 506 1056 577 1.75 1.96 3.75 -7.57 12.66 -0.67 -1.54 0.
Car 0.00 0 -1.57 1072 494 1124 552 1.75 1.96 3.75 -7.57 12.66 -0.67 -1.54 0.

VOC

  • 目录

    标注文件目录结构

    +--- Annotations
|   +--- 000000000.xml
|   +--- 000000001.xml
+--- JPEGImages
|   +--- 000000000.jpg
|   +--- 000000001.jpg

    推理文件目录结构

    +--- 000000000.xml
+--- 000000001.xml
  • 示例标注/推理文件 推理含score字段

    2D目标检测-VOC.xml

    <?xml version="1.0" encoding="utf-8"?>
<annotation>
    <size>
        <depth>3</depth>
        <width>1920</width>
        <height>1020</height>
    </size>
    <label_counts>
        <label_num>1</label_num>
        <label_meta_id>373</label_meta_id>
    </label_counts>
    <label_counts>
        <label_num>12</label_num>
        <label_meta_id>372</label_meta_id>
    </label_counts>
    <segmented>0</segmented>
    <folder>dataset-1064version-1</folder>
    <label_task_id>14224</label_task_id>
    <origin_name>1598929950099.jpg</origin_name>
    <object>
        <label_meta_id>372</label_meta_id>
        <occluded>0</occluded>
        <bndbox>
            <ymin>503</ymin>
            <xmax>980</xmax>
            <ymax>524</ymax>
            <xmin>956</xmin>
        </bndbox>
        <pose>Unspecified</pose>
        <attribute>{"遮挡":"0","截断":"0"}</attribute>
        <truncated>0</truncated>
        <difficult>1</difficult>
        <name>Car</name>
        <serial_number>16</serial_number>
        <shape_type>bndbox</shape_type>
    </object>
    <filename>hash0-1598929950099.xml</filename>
</annotation>
父主题： 模型数据集支持
分享：

    相关文档

    相关产品