Device, method, and program for detecting boundary lines included in captured images
AssigneeMobility Technologies
InventorHiroto Honda
Life of this patent
Estimating the term…
Abstract
A CPU 11 of a boundary line detection device 10 according to an embodiment of the present invention is configured to function as an image acquisition component 112 that acquires images captured (inputted) via an image input device (a camera or the like), captured images received via a network, and so forth, and a boundary line detector 114 that detects a boundary line included in these captured images. The boundary line detector 114 is configured to specify, using a learned model 152, a plurality of estimated points estimated to be points on a boundary line included in a captured image and to detect the boundary line included in the captured image on the basis of at least the specified estimated points.
Claims11
A device for detecting a boundary line included in a captured image, comprising one or more computer processors,
- wherein, in response to the execution of a readable instruction, said one or more computer processors executes:
- acquiring a captured image and
- detecting a boundary line included in the acquired captured image;
- and
- the detection includes specifying, using a learned model, a plurality of estimated points estimated to be points on a boundary line included in the acquired captured image and detecting the boundary line included in the acquired captured image based on at least the specified plurality of estimated points,
- wherein the learned model is generated by machine learning that uses, as learning data, a plurality of learning images each of which is associated with information including first position specifying information that specifies a position in the width direction at a first height of a boundary line included in the image and second position specifying information that specifies a position in the width direction at a second height of a boundary line included in the image, and is configured so that first and second position specifying information corresponding to an inputted image are estimated and outputted, and
- the detection includes specifying the plurality of estimated points based n the first and second position specifying information outputted from the learned model.
A method for detecting a boundary line included in a captured image, executed by one or more computers,
- wherein said method comprises:
- acquiring a captured image;
- and
- detecting a boundary line included in the acquired captured image;
- and
- the detection includes specifying, using a learned model, a plurality of estimated points estimated to be points on a boundary line included in the acquired captured image and detecting the boundary line included in the acquired captured image based on at least the specified plurality of estimated points,
- wherein the learned model is generated by machine learning that uses, as learning data, a plurality of learning images each of which is associated with information including first position specifying information that specifies a position in the width direction at a first height of a boundary line included in the image and second position specifying information that specifies a position in the width direction at a second height of a boundary line included in the image, and is configured so that first and second position specifying information corresponding to an inputted image are estimated and outputted, and
- the detection includes specifying the plurality of estimated points based on the first and second position specifying information outputted from the learned model.
A non-transitory computer-readable medium including instructions for detecting a boundary line included in a captured image, the instructions to be performed on a processor, the instructions, comprising:
- processing to acquire a captured image and
- processing to detect a boundary line included in the acquired captured image;
- and
- the detection processing includes specifying, using a learned model, a plurality of estimated points estimated to be points on a boundary line included in the acquired captured image and detecting the boundary line included in the acquired captured image based on at least the specified plurality of estimated points,
- wherein the detection includes distinguishably specifying a first type of estimated points estimated to be located on a first type of boundary line and a second type of estimated points estimated to be located on a second type of boundary line, and distinguishably detecting the first type of boundary line and the second type of boundary line based on at least the specified plurality of first and second types of estimated points, and
- wherein the learned model is generated by machine learning that uses, as learning data, a plurality of learning images each of which is associated with information including third position specifying information that specifies the position of points on a first type of boundary line included in the image and fourth position specifying information that specifies the position of points on a second type of boundary line included in the image, and is configured so that third and fourth position specifying information corresponding to an inputted image are estimated and outputted, and
- the detection includes specifying a first of estimated points based on third position specifying information outputted from the specific learned model, while also specifying a second type of estimated points based on fourth position specifying information outputted from the specific learned model.
Dependencies are inferred from the claim text.