knn分类器根据Features计算得到分类和置信度报错

请问大佬，我在有SelectedImage1的情况下，再经过calculate_features ()后得到Features，但是再将运行classify_class_knn (KNNHandle, Features, Class, Rating)时报错Wrong number of values of control parameter 2(HALCON错误代码:1402)

此时Features的值为[0.1, 3.23, 3.78, 5.2, 6.34, 8.81, 17.23, 18.83, 31.09, 42.79, 34.35, 30.26, 34.74, 32.35, 26.06, 26.05, 26.19, 26.93, 23.53, 18.86, 16.9, 16.43, 16.87, 16.49, 17.48, 23.02, 26.11, 27.23, 47.52, 82.18, 105.1, 90.63, 70.43, 64.4, 64.02, 59.27, 47.79, 44.51, 43.16, 26.45]

请问大佬这是为什么呀，按道理讲KNNHandle没问题，Features没问题，输出的Class和Rating

1. binary_threshold (SelectedImage1, Region, 'max_separability', 'dark', UsedThreshold)
   
2. * The objects are separated before calculating the features
   
3. * for each object sample
   
4. connection (Region, ConnectedRegions)
   
5. select_shape (ConnectedRegions, SelectedRegions1, 'area', 'and', 62746.4, 100000)
   
6. * count_obj (SelectedRegions1, Number)
   
7. * Extract the selected features
   
8. calculate_features (SelectedRegions1, SelectedImage1, SelectedFeatures, Features)

复制代码

Sele

救命啊 有没有人救救我

有完整代码么，用你提供的示例，SelectedRegions1为空

Lin_She 发表于 2024-10-22 15:49
有完整代码么，用你提供的示例，SelectedRegions1为空

训练时用的是

1. select_feature_set_knn (ClassTrainDataHandle, 'greedy', [], [], KNNHandle, SelectedFeatures, Score)

复制代码

这里得出的SelectedFeatures是

1. ['gray_hor_proj', 'gray_rb', 'polar_gray_proj_histo', 'gray_hor_proj_histo']

复制代码

，但是我在代码中想要指定features，
然后我将这里的KNNHandle保存在了文件knn_model_file.hknn中
然后下面就是测试的过程

1. read_class_knn('knn_model_file.hknn', KNNHandle)
   
2. * 这里是我指定的features
   
3. SelectedFeatures:=['polar_gray_proj_histo', 'gray_rb', 'cooc_pyramid_4', 'gray_vert_proj_histo', 'edge_density_pyramid_4', 'edge_density_pyramid_2']
   
4. dev_display (SelectedImage1)
   
5. * Segment objects from the image
   
6. binary_threshold (SelectedImage1, Region, 'max_separability', 'dark', UsedThreshold)
   
7. * The objects are separated before calculating the features
   
8. * for each object sample
   
9. connection (Region, ConnectedRegions)
   
10. select_shape (ConnectedRegions, SelectedRegions1, 'area', 'and', 67035.1, 200000)
    
11. * count_obj (SelectedRegions1, Number)
    
12. * Extract the selected features
    
13. calculate_features (SelectedRegions1, SelectedImage1, SelectedFeatures, Features)
    
14. * Classify object with the selected features
    
15. * and the automatically created classifier
    
16. classify_class_knn (KNNHandle, Features, Class, Rating)

复制代码

这里就会在输入KNNHanle, Features之后，得到了Class=0，但是Rating为空，并且报错Wrong number of values of control parameter 2(HALCON错误代码:1402)

我理解是select_feature_set_knn输出的KNNHandle只能使用SelectedFeatures输出的参数，如果你想自己指定参数，可能你需要直接保存ClassTrainDataHandle。
select_feature_set_knn

功能：从一组要素中选择最佳子集以解决某个分类问题。

尝试一下

Lin_She 发表于 2024-10-23 10:38
我理解是select_feature_set_knn输出的KNNHandle只能使用SelectedFeatures输出的参数，如果你想自己指定参 ...

说错了，不是保存训练数据句柄，是不要进行select_feature_set_knn，直接保存KNNHandle试试

学习一下