Method and device for improving object recognition rate of self-driving car
Assignee42Dot Inc.
InventorsMyoung Hoon Cho, Jae Yun Lee, Phuoc Tien Nguyen
Life of this patent
Estimating the term…
Abstract
Provided is a method of improving an object recognition rate including recognizing a first object in a first image obtained while driving, detecting whether the recognized first object has disappeared for a preset time period and then reappeared in the first image, based on detecting that the first object has reappeared, calculating training data for the first object, and controlling such that a recognition model for recognizing an object included in an image is to be trained based on information based on the calculated training data.
Claims13
A method of improving an object recognition rate of a self-driving car, the method comprising:
- recognizing a first object in a first image obtained while driving;
- detecting whether the recognized first object has disappeared for a preset time period and then reappeared in the first image;
- based on detecting that the first object has reappeared, calculating training data for the first object;
- controlling such that a recognition model for recognizing an object included in an image is to be trained based on information based on the calculated training data;
- and
- calculating information based on the calculated training data by filtering the calculated training data based on a preset filtering criterion,
- wherein the first image is a video comprising a plurality of frames,
- the preset filtering criterion is a filtering criterion for a time length of a series of frames in which the first object has disappeared for the preset time period in a second frame after being recognized in a first frame and then reappeared in a third frame,
- the calculating of the information based on the calculated training data comprises, based on a time length between the first frame and the third frame being longer than a reference frame length, calculating the information based on the training data, and
- the reference frame length is determined according to at least one of a frame rate of the first image, a moving speed of the first object in the first image, a field of view of a camera that has captured the first image, an angle of the camera that has captured the first image, and a distortion rate of a lens of the camera that has captured the first image.
A device for improving an object recognition rate of a self-driving car, the device comprising:
- a memory storing at least one program;
- and
- a processor configured to perform an operation by executing the at least one program,
- wherein the processor is further configured to:
- recognize a first object in a first image obtained while driving,
- detect whether the recognized first object has disappeared for a preset time period and then reappeared in the first image,
- based on detecting that the first object has reappeared, calculate training data for the first object,
- control a recognition model for recognizing an object included in an image to be trained based on information based on the calculated training data, and
- calculate information based on the calculated training data by filtering the calculated training data based on a preset filtering criterion,
- wherein the first image is a video comprising a plurality of frames,
- the preset filtering criterion is a filtering criterion for a time length of a series of frames in which the first object has disappeared for the preset time period in a second frame after being recognized in a first frame and then reappeared in a third frame,
- wherein the processor is configured to calculate the information based on the calculated training data, further based on a time length between the first frame and the third frame being longer than a reference frame length,
- wherein the reference frame length is determined according to at least one of a frame rate of the first image, a moving speed of the first object in the first image, a field of view of a camera that has captured the first image, an angle of the camera that has captured the first image, and a distortion rate of a lens of the camera that has captured the first image.
Dependencies are inferred from the claim text.