Skip to content

测试rknn分类模型与目标检测模型指标

Notifications You must be signed in to change notification settings

Uhao-P/rk3588_metrics

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

1 Commit
 
 
 
 
 
 
 
 
 
 

Repository files navigation

提供分类模型与目标检测模型的指标评估

分类模型目录结构

  • data
    • 0
      • **.jpg
      • **.jpg
    • 1
      • **.jpg
      • **.jpg
    • 2
      • **.jpg
      • **.jpg
    • ...

检测模型目录结构

  • data
    • imgs
      • 00001.jpg
      • 00002.jpg
      • 00003.jpg
      • 00004.jpg
      • ...
    • detections
      • 00001.txt
      • 00002.txt
      • 00003.txt
      • 00004.txt
      • ...
    • groundtruths
      • 00001.txt
      • 00002.txt
      • 00003.txt
      • 00004.txt
      • ...
    • res
      • 00001.jpg
      • 00002.jpg
      • 00003.jpg
      • 00004.jpg
      • ...

detections文件夹下’00001.txt‘内容示例

# idClass  confidence  x  y  w  h
object .88 5 67 31 48
object .70 119 111 40 67
object .80 124 9 49 67

groundtruths文件夹下’00001.txt‘内容示例

# idClass  x  y  w  h
object 25 16 38 56
object 129 123 41 62

分类模型准确率评估

需要按照上面介绍的‘分类模型目录结构’规范要验证的数据

./classification model_path dataset_path

检测模型准确率评估(以yolov5/7为例)

需要按照上面介绍的‘分类模型目录结构’规范要验证的数据

./detection model_path dataset_path

接下来请参考https://github.com/rafaelpadilla/Object-Detection-Metrics 中计算方式

About

测试rknn分类模型与目标检测模型指标

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published