Endoscopic image learning device, endoscopic image learning method, endoscopic image learning program, and endoscopic image recognition device
An object is to provide an endoscopic image learning device, an endoscopic image learning method, an endoscopic image learning program, and an endoscopic image recognition device that appropriately learn a learning model for image recognition for recognizing an endoscopic image in which a treatment tool for an endoscope appears. The object is achieved by an endoscopic image learning device including an image generation unit and a machine learning unit. The image generation unit generates a superimposed image where a foreground image in which a treatment tool for an endoscope is extracted is superimposed on a background-endoscopic image serving as a background of the foreground image, and the machine learning unit performs the learning of a learning model for image recognition using the superimposed image.
1. An endoscopic image learning device comprising:
an image generation unit that generates a superimposed image where a foreground image in which an image of a treatment tool disposed on a distal end of an endoscope is extracted is superimposed on a background-endoscopic image serving as a background of the foreground image; and
a machine learning unit that performs learning of a learning model for image recognition using the superimposed image.
2. The endoscopic image learning device according to claim 1 , further comprising:
a foreground image acquisition unit that acquires the foreground image; and
a background image acquisition unit that acquires the background-endoscopic image.
3. The endoscopic image learning device according to claim 1 , further comprising:
a learning-endoscopic image acquisition unit that acquires a learning-endoscopic image,
wherein the machine learning unit performs the learning using the learning-endoscopic image.
4. The endoscopic image learning device according to claim 1 , further comprising:
an image processing section that performs specific processing on the foreground image.
5. The endoscopic image learning device according to claim 4 ,
wherein the specific processing is at least one of affine transformation processing, color conversion processing, or noise application processing.
6. The endoscopic image learning device according to claim 1 , further comprising:
a foreground-material image acquisition section that acquires a foreground-material image including the treatment tool for an endoscope; and
an image cut-out section that cuts out the treatment tool for an endoscope from the foreground-material image to generate the foreground image.
7. The endoscopic image learning device according to claim 6 ,
wherein the foreground-material image is an endoscopic image that is picked up in a case where the treatment tool for an endoscope is used in an endoscope apparatus.
8. The endoscopic image learning device according to claim 6 ,
wherein the foreground-material image is an image other than an endoscopic image.
9. The endoscopic image learning device according to claim 1 ,
wherein the learning model is at least one of a learning model for recognizing whether or not the treatment tool for an endoscope is present, a learning model for recognizing a type of the treatment tool for an endoscope, a learning model for distinguishing a region of the treatment tool for an endoscope from a region other than the region of the treatment tool for an endoscope, a learning model for detecting a position of a region of interest, or a learning model for classifying an image.
10. The endoscopic image learning device according to claim 1 ,
wherein the machine learning unit performs the learning using a convolution neural network.
11. An endoscopic image recognition device comprising:
an image acquisition unit that acquires an endoscopic image; and
an image recognition unit that recognizes the endoscopic image acquired by the image acquisition unit by using a learning model learned by the endoscopic image learning device according to claim 1 .
12. An endoscopic image learning method comprising:
generating a superimposed image where a foreground image in which an image of a treatment tool disposed on a distal end of an endoscope is extracted is superimposed on a background-endoscopic image serving as a background of the foreground image; and
performing learning of a learning model for image recognition using the superimposed image.
13. A non-transitory, tangible recording medium which records thereon a program that, when executed by a computer, causes the computer to perform an endoscopic image learning method including:
generating a superimposed image where a foreground image in which an image of a treatment tool disposed on a distal end of an endoscope is extracted is superimposed on a background-endoscopic image serving as a background of the foreground image; and
performing learning of a learning model for image recognition using the superimposed image.