Vehicle body information determination device, vehicle body information determination method, and program

Abstract

This technology provides the ability to quickly detect target vehicles. [Solution] The vehicle information determination device (2) includes: an extraction processing unit (23) that extracts an image for identifying the vehicle type from an image of a photographed vehicle; an identification unit (24) that identifies the vehicle type from the image extracted by the extraction processing unit (23) using a learning model that has learned the relationship between a learning image and a label for identifying the vehicle type; and a vehicle color determination unit (25) that determines the vehicle body color based on the image extracted by the extraction processing unit (23) and the vehicle type identified by the identification unit (24).

Claims

1. A cropping processing unit that extracts an image from a photographed image of a car to identify the make and model of the car, An identification unit identifies the type of vehicle from the image extracted by the extraction processing unit, using a learning model that has learned the relationship between training images and labels for identifying the type of vehicle of the vehicle. A vehicle body information determination device comprising: a determination unit that determines the body color of the automobile based on the image extracted by the extraction unit and the type of automobile identified by the identification unit.

2. The vehicle body information determination device according to claim 1, wherein the determination unit determines the vehicle body color by referring to an image of a vehicle body color determination region that is set in advance according to the type of vehicle in the image extracted by the extraction processing unit.

3. The vehicle body information determination device further includes a storage unit that stores characteristic quantities of color components corresponding to multiple candidate colors of the vehicle body set for each type of automobile, The vehicle body information determination device according to claim 2, wherein the determination unit calculates a similarity score based on the feature quantities for each color component of each candidate color and the feature quantities for each color component of the image of the vehicle body color determination area, and determines which of the plurality of candidate colors the vehicle body color corresponds to based on the similarity score.

4. The vehicle body information determination device according to claim 3, wherein the feature quantities for each color component are histogram feature quantities for each color component of each pixel in the image of the vehicle body color determination region.

5. The vehicle body information determination device according to claim 3 or 4, wherein the determination unit calculates a Bhattacharya distance between the histogram feature quantity of the color component of the candidate color and the histogram feature quantity of the color component of the image of the vehicle body color determination region for each color component, and calculates the similarity score by performing weighted addition to the Bhattacharya distance for each color component.

6. The process involves extracting an image from a photograph of a vehicle to identify the vehicle model, The steps include identifying the vehicle type from the extracted image using a learning model that has learned the relationship between training images and labels for identifying the vehicle type, A method for determining vehicle body information, comprising the step of determining the body color of the automobile based on the cropped image and the type of automobile.

7. On the computer, The process involves extracting an image from a photograph of a car to identify the make and model of that car, Using a learning model that has learned the relationship between training images and labels for identifying the vehicle type, the process of identifying the vehicle type from the extracted image is performed. A program that performs a process to determine the body color of a car based on the extracted image and the car model.

Images