Image classification device and image classification method
Provided are an image classification device and method that are capable of extracting and mapping an important feature in an image. The image classification device includes: a feature extraction unit 101 that generates a first image group generated by applying different noises to the same image among images included in an image group and a second image group including different images, is trained such that features obtained from the first image group are approximate, is trained such that features obtained from the second image group are more different, and extracts features; a feature mapping unit 102 that maps the extracted plurality of features two-dimensionally or three-dimensionally using manifold learning; and a display unit 103 that displays a mapping result and constructs a training information application task screen.
1 . An image classification device for performing image classification, the image classification device comprising:
a feature extraction unit that generates a first image group generated by applying different noises to the same image among images included in an image group and a second image group including different images, that is trained such that features obtained from the first image group are approximate, that is trained such that features obtained from the second image group are more different, and that extracts features;
a feature mapping unit that maps the extracted plurality of features two-dimensionally or three-dimensionally using manifold learning; and
a display unit that displays a training information application task screen on which a mapping result is displayed and a training information application task is performed.
2 . The image classification device according to claim 1 , wherein
the feature extraction unit includes a dimension reduction unit that is trained using training information in order to avoid deterioration in mapping accuracy due to the manifold learning when the number of dimensions of the extracted features is large.
3 . The image classification device according to claim 1 , wherein
on the training information application task screen, a plurality of features included in a specified region are remapped and displayed.
4 . The image classification device according to claim 1 , wherein
on the training information application task screen, a feature having a long distance from a neighboring point among the mapped plurality of features is emphasized and displayed.
5 . The image classification device according to claim 1 , wherein
on the training information application task screen, a region in which a distribution density of features among the mapped plurality of features is high is emphasized and displayed.
6 . The image classification device according to claim 1 , wherein
on the training information application task screen, image thumbnails are displayed at positions where the plurality of features are mapped.
7 . The image classification device according to claim 1 , wherein
on the training information application task screen, training information is to be applied to the mapped plurality of features at once by selecting a region on a map.
8 . The image classification device according to claim 1 , wherein
on the training information application task screen, a clustering result of the plurality of features output by the feature extraction unit and a clustering result using two-dimensional coordinates or three-dimensional coordinates output by the feature mapping unit are compared, and a region to be remapped is displayed when the results are different.
9 . An image classification method for performing image classification, the image classification method comprising:
generating a first image group generated by applying different noises to the same image among images included in an image group and a second image group including different images, performing training such that features obtained from the first image group are approximate, performing training such that features obtained from the second image group are more different, and extracting features;
mapping the extracted plurality of features two-dimensionally or three-dimensionally using manifold learning; and
displaying a mapping result on a training information application task screen.
10 . The image classification method according to claim 9 , wherein
dimensions are reduced and output in order to avoid deterioration in mapping accuracy due to the manifold learning when the number of dimensions of the extracted features is large.
11 . The image classification method according to claim 9 , wherein
on the training information application task screen, a clustering result of the plurality of features and a clustering result using two-dimensional coordinates or three-dimensional coordinates are compared, and a region to be remapped is displayed when the results are different.