Denso had four patents in artificial intelligence during Q3 2023. One patent is for an on-vehicle device that uploads information about a vehicle to an external device, including trimming information from captured images. Another patent is for a system that controls a robot arm to pour fluid from one container to another, using machine learning to generate a learning model. Another patent is for a system that produces a data set to train a neural network for object detection, comparing different sets of objects detected by different systems. Another patent is for a system and method to determine the location of a portable device relative to an object. Lastly, there is a patent for an image recognition apparatus that uses multiple object detection processes with different models to generate inference result data based on the degree of overlap among detected object regions. GlobalData’s report on Denso gives a 360-degreee view of the company including its patenting strategy. Buy the report here.
Denso grant share with artificial intelligence as a theme is 50% in Q3 2023. Grant share is based on the ratio of number of grants to total number of patents.
Application: On-vehicle device, management system, and upload method (Patent ID: US20230276059A1)
The patent filed by Denso Corp. describes an on-vehicle device that uploads information about a vehicle to an external device at regular intervals. The device includes a processor that uploads trimming information, which involves removing a specific area from the captured image information.
Claim 1 states that the on-vehicle device comprises a processor that generates a compressed file from the captured image information. This compressed file includes a first frame that contains all the information in a frame, as well as multiple second frames that represent the difference between a frame and the first frame or the previous frame. The processor also uploads trimming information to an external device, where a target area is trimmed from the first frame in the compressed file.
Claim 2 adds that the processor can upload the trimming information within a specified time period requested by the external device or the image information before trimming, depending on the request made by the external device.
Claim 3 explains that the processor can identify the target area where a specific object, specified by an image recognition process, appears in the image information. It then generates the trimming information based on this identified target area.
Claim 4 mentions that the processor can exclude information related to areas other than the target area from the first frame to generate the trimming information.
Claim 5 introduces a management system that includes the on-vehicle device described in claim 1. This system also includes a management device that aggregates and manages the information uploaded from the on-vehicle device.
Claim 6 describes an upload method that involves generating a compressed file from the captured image information. This compressed file includes a first frame that holds all the information in a frame, as well as multiple second frames representing the difference between a frame and the first frame or the previous frame. The method also includes uploading trimming information to an external device, where a target area is trimmed from the first frame in the compressed file.
In summary, the patent describes an on-vehicle device that uploads vehicle information and includes a processor for generating compressed files and uploading trimming information. The device can identify and trim specific target areas from the captured image information. The patent also introduces a management system that includes the on-vehicle device and a management device for aggregating and managing the uploaded information.