CN-122003704-A - Image storage device for vehicle and operation method thereof

Abstract

The invention discloses an image storage device which can be installed on a vehicle and can shoot and store images around the vehicle. The image storage device comprises a shooting part for shooting and acquiring a real-time running image of a vehicle, an accident prediction part for analyzing the real-time running image by utilizing an accident detection artificial intelligent model learned in advance and predicting the accident occurrence probability of the vehicle, a storage part for storing the running image of a preset time period before and after the accident detection moment according to the accident detection moment when judging that the accident occurrence probability exists, and a communication part for transmitting the running image of the preset time period to an error ratio judging server, wherein the accident detection artificial intelligent model is a model trained by utilizing training data marked with the accident image and a normal running image in a supervision learning mode.

Inventors

• CAI ZHENHAO

Assignees

• 恩迪人工智能株式会社

Dates

Publication Date

20260508

Application Date

20250305

Priority Date

20240904

Claims (10)

1. 1. An image storage device that can be mounted on a vehicle and that captures and stores images around the vehicle, comprising: A shooting part for shooting and acquiring real-time running images of the vehicle; an accident prediction unit that analyzes the real-time traveling image using a pre-learned accident detection artificial intelligence model, and predicts an accident occurrence probability of the vehicle; A storage unit for storing a travel image including a predetermined time period before and after the accident detection time, based on the accident detection time, when it is determined that the possibility of occurrence of the accident exists, and A communication unit for transmitting the running image of the predetermined period to a loss ratio determination server, The accident detection artificial intelligent model is a model which is trained by using training data marked with accident images and normal driving images and by a supervised learning mode.
2. 2. The image storage device of claim 1, wherein, The accident detection artificial intelligence model extracts and analyzes characteristics of the running image according to each frame or a preset number of frame group units of the real-time running image to predict the accident occurrence probability.
3. 3. The image storage device according to claim 1, wherein the accident detection artificial intelligence model is a model that is learned by transfer learning in which weight values of a model that is learned in advance are fine-tuned by using the accident image and a normal running image.
4. 4. The image storage device of claim 1, wherein, The sensor part comprises at least one sensor of a gyro sensor, an acceleration sensor, an impact sensor, a global positioning system and a laser radar sensor, The accident prediction unit determines that there is a possibility of occurrence of an accident in the vehicle when a combination index obtained by combining detection information obtained by at least one sensor included in the sensor unit and accident occurrence possibility information determined by the accident detection artificial intelligence model is equal to or more than a preset threshold.
5. 5. The image storage device of claim 4, wherein, When the accident occurrence probability information determined by the accident detection artificial intelligence model is p (p is a real number between 0 and 1), the distance between the vehicle and the surrounding objects measured by the lidar sensor is D, α and β are predetermined weight values, and ε is a predetermined constant value, the combination index (C) is determined according to the following mathematical formula But is determined in a manner that is proportional to the probability of occurrence of an accident determined by the accident detection artificial intelligence model and inversely proportional to the distance between the vehicle and surrounding objects, When the combination index (C) is equal to or greater than the preset threshold value, the accident prediction unit determines that there is a possibility of accident occurrence between the vehicle and the surrounding objects.
6. 6. The image storage device of claim 1, wherein, The transient ratio determination server identifies a main object for predicting a transient ratio from the running image using a pre-learned object identification artificial intelligent model, and determines a transient ratio between the vehicle and the main object from the running image of a predetermined period of time transmitted from the image storage device using a pre-learned transient ratio determination artificial intelligent model.
7. 7. The image storage device of claim 6, wherein, The object recognition artificial intelligence model extracts the spatio-temporal characteristics of the driving image of the predetermined period of time frame by frame, tracks and analyzes the motion and scene changes of the object in units of frames, The passing ratio judgment artificial intelligence model is a model trained in a supervised learning or unsupervised learning manner on training data, which is labeled based on a basis of a basic passing ratio for classifying a running image preprocessed by the accident detection artificial intelligence model in a passing ratio information portal.
8. 8. A method of operating a vehicular image storage device that can be mounted on a vehicle and that captures and stores images around the vehicle, comprising the steps of: shooting and acquiring a real-time driving image of the vehicle; analyzing the real-time driving image by utilizing a pre-learned accident detection artificial intelligent model, and predicting the accident occurrence probability of the vehicle; storing a driving image including a predetermined period of time before and after the accident detection time, based on the accident detection time, when it is determined that the possibility of the accident occurs; transmitting the running image of the predetermined period to a loss ratio judging server, The accident detection artificial intelligent model is trained by a supervised learning mode by utilizing training data marked with accident images and normal driving images.
9. 9. An image storage device capable of being mounted on a first vehicle and capturing and storing images around the vehicle, comprising: a photographing part for photographing and acquiring a real-time driving image of the first vehicle; An accident prediction unit that analyzes the real-time traveling image using a pre-learned accident detection artificial intelligence model, and predicts the probability of occurrence of an accident for a second vehicle around the first vehicle; a storage unit for storing a travel image including a predetermined time period before and after the accident detection time, based on the accident detection time, when it is determined that the possibility of occurrence of the accident is equal to or greater than a predetermined threshold value, and A communication unit that transmits the running video for the predetermined period to a loss ratio determination server and/or a terminal of a user of the second vehicle, The accident detection artificial intelligent model is trained by a supervised learning mode by utilizing training data marked with accident images and normal driving images.
10. 10. A method of operating an image storage device that is mountable on a first vehicle and that captures and stores images of the surroundings of the vehicle, comprising the steps of: Shooting and acquiring a real-time driving image of the first vehicle; Analyzing the real-time driving image by utilizing a pre-learned accident detection artificial intelligent model, and predicting the accident occurrence probability of a second vehicle around the first vehicle; Storing a running image including a predetermined period of time before and after the accident detection time, based on the accident detection time, when it is determined that the possibility of occurrence of the accident is equal to or greater than a predetermined threshold value, and Transmitting the running image of the predetermined period to a loss ratio judging server and/or a terminal of a user of the second vehicle, The accident detection artificial intelligent model is trained by a supervised learning mode by utilizing training data marked with accident images and normal driving images.

Description

Image storage device for vehicle and operation method thereof Technical Field The present disclosure relates to a black box for a vehicle, and more particularly, to an image storage device for a vehicle, which has an on-device (on-device) artificial intelligence model, and predicts an accident occurrence from a real-time driving image using the artificial intelligence model, and stores an accident image. The present disclosure also relates to a failure rate determination system that transmits an accident-related image captured by a vehicle black box to a server, and determines a failure rate in the accident image by the server. Background A black box for a vehicle is a device attached to a vehicle for recording an accident or situation occurring during running, and is intended to analyze the cause of the accident and track responsibility. The black box for a vehicle is also called a Dashboard Camera (Dashcam) or a drive recorder (Dashcam). The black box for a vehicle is used to analyze the cause of an accident and to track down responsibility, but often stores unnecessary images instead of direct images of the accident, which may cause a shortage of memory or battery, thereby failing to secure the images required at the time of the accident. For example, in a conventional automobile data recorder, an image is stored by motion recognition or impact detection, but the motion or impact recognized by the black box is often not related to an accident occurring in an actual vehicle. Also, if a problem occurs in the built-in battery of the black box, the stored image date may be initialized or not stored, thereby causing a case where the related image cannot be secured in the event of an accident. Further, if the storage capacity of the black box is insufficient, the image may be stored so as to cover the previously stored image data, and thus the accident-related image may not be ensured. In addition, because the process of extracting and transmitting the accident image stored in the memory is complex, when a driver submits the image data to an insurance company, the driver often cannot submit the original image, but uses a personal smart phone to shoot the picture of the black box for submitting, and the indirectly shot image may obstruct the accuracy of the error judgment in the insurance processing process. Disclosure of Invention An object of the present disclosure is to provide an image storage device for a vehicle that stores a related travel image when a traffic accident is predicted using an accident prediction artificial intelligence model. Further, an object of the present disclosure is to provide a system for transmitting an accident image captured by a vehicle image storage device to an external server, so that a user or a stakeholder can share the accident image. Further, it is an object of the present disclosure to provide a method and apparatus for determining a contrast ratio between related objects from an image using a contrast ratio determination artificial intelligence model. An image storage device for a vehicle according to one embodiment of the present disclosure includes an imaging unit that captures and acquires a real-time running image of a vehicle, an accident prediction unit that analyzes the real-time running image using an accident detection artificial intelligence model learned in advance, predicts an accident occurrence probability of the vehicle, a storage unit that stores a running image including a predetermined period of time before and after the accident detection time, based on the accident detection time, when it is determined that the accident occurrence probability exists, and a communication unit that transmits the running image of the predetermined period of time to an error rate determination server, the accident detection artificial intelligence model being a model trained by a supervised learning method using training data labeled with the accident image and the normal running image. According to the embodiment of the disclosure, the artificial intelligence accident detection model stores images only when traffic accidents occur, so that blind storage is avoided, existing images do not need to be covered, the images can be safely stored without affecting the limited capacity and service life of a storage space, and the risk of image loss does not exist. In addition, according to the embodiment of the disclosure, the stored image is transmitted through the connection server, so that personal information contained in the accident image can be shielded, the passing rate between accident vehicles can be judged, and when a user requests, data in the form of a webpage or an application program including insurance acceptance can be provided through the platform. Drawings Fig. 1 is a diagram showing an overall system of a vehicle black box for determining a failure rate by analyzing an accident image while sharing an image captured by a vehicle image storage device according