Annotation assisting method, annotation assisting device, and recording medium
An annotation, memory technology, applied in instrumentation, calculation, character and pattern recognition, etc., can solve problems such as tasks that do not consider the identification of measurement elements, etc.
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
Method used
Image
Examples
no. 1 approach
[0059] figure 1 It is a block diagram showing an example of the overall configuration of the annotation system according to the first embodiment of the present invention. The annotation system is a system for assisting a worker in annotating work of setting annotation data on image data. The annotation (annotation) work is a work performed by a worker to generate learning data used when machine learning a classifier that is a learning target. The operator finds out the recognition object of the predetermined classification from the image presented on the display, sets the area information indicating the area where the recognition object appears, and inputs the classification mark indicating the classification of the object to the image. Data sets the annotation data.
[0060] Recognition objects are, for example, people, ordinary cars, trucks, bicycles, motorcycles, and the like. Classification is to identify the type of object. The category mark is text data representing ...
no. 2 approach
[0150] The second embodiment is an embodiment for determining the optimum value of the first threshold value for the reliability used to discriminate whether to let the object recognizer set the pre-annotated data on the image data. In addition, in this embodiment, the same code|symbol is attached|subjected to the same component as 1st Embodiment, and description is abbreviate|omitted. Furthermore, in the second embodiment, since the overall configuration of the annotation system is the same as figure 1 Similarly, the configuration of the server 100 is the same as figure 2 same, so use figure 1 and figure 2 Be explained. This also applies to the embodiments described later.
[0151] When the value of the reliability of the recognition result output by the recognizer together with the recognition result is equal to or greater than the first threshold, prior annotation data is set for the recognition target object.
[0152] refer to figure 2 . In this embodiment, the s...
no. 3 approach
[0179] The third embodiment is an embodiment for determining the optimum value of the second threshold value information related to the area size indicated by the area information.
[0180] The object recognizer, in the case of recognizing the recognition object based on the image data, if the region size shown in the region information set for the recognized recognition object is within the lower limit threshold and the upper limit threshold included in the second threshold value information Within the range, the area information is set. If the size of the objects appearing on the image data is too small or too large, the recognition accuracy of the object recognizer will be reduced. Therefore, by setting a threshold value for the area size, it is possible to suppress the occurrence of the object recognizer erroneously detecting the recognized object.
[0181] In addition, the area size may represent at least one of the vertical width, the horizontal width, and the area of t...
PUM
Login to View More Abstract
Description
Claims
Application Information
Login to View More