Driving recorder and image storage method

Abstract

A driving recorder is suitable for a remote server and comprises an image acquisition module, a communication module, a first storage device and a processing module. The processing module is electrically connected with the image acquisition module, the communication module and the first storage device, and executes an image storage method so as to transmit a preset interval time-lapse image of a driving image to the remote server for storage through the communication module at least when a current weather state meets a weather preset condition; therefore, storage of all images can be avoided, bandwidth is saved, and the driving image is effectively stored.

Description

Technical Field

[0001] The present invention relates to a driving recorder and an image saving method, and more particularly to a driving recorder and an image saving method capable of saving bandwidth and effectively saving driving images. Background Art

[0002] A driving recorder is a video recording device installed on a vehicle to record driving images (especially front images) after activation, so that when a traffic accident occurs, the stored driving images can be used as the most powerful evidence. Conventional driving recorders usually use an external memory card (such as an SD card) as a storage medium. However, in a traffic accident, if the impact force is too large or a burn occurs, it is likely to cause the memory card in the driving recorder to be damaged. Especially in bad weather, the probability and severity of traffic accidents are relatively increased due to factors such as road conditions and poor visibility, and the possibility of accidental damage to the memory card in the driving recorder is also increased, causing the driving images recorded during the accident to disappear. Therefore, another conventional method uses a driving recorder with a networking function to establish a connection with a remote server, so that the driving recorder can continuously upload the recorded driving images to the remote server, so that the memory card can upload the recorded driving images before it is damaged and keep them. However, due to the huge amount of driving image data, this conventional method must also have sufficient bandwidth. Therefore, whether there are other driving recorders that can effectively save driving images has become a problem to be solved. Summary of the invention

[0003] Therefore, an object of the present invention is to provide a driving recorder that can save bandwidth and effectively store driving images.

[0004] Therefore, one aspect of the present invention is to provide a driving recorder, which is applicable to a remote server and includes an image capture module, a communication module, a first storage device, and a processing module. The image capture module is used to capture a driving image. The communication module is used to provide a networking function. The first storage device is used to store the driving image.

[0005] The processing module is electrically connected to the image capture module, the communication module, and the first storage device, and transmits a predetermined interval of the driving image through the communication module to the remote server for storage when the current weather state meets the predetermined weather condition.

[0006] In some embodiments, the processing module inputs multiple frames of the driving image into a deep learning model to obtain the weather condition corresponding to a judgment score. When the processing module determines that the judgment scores obtained continuously by the deep learning model are gradually approaching the judgment score corresponding to a bad weather level and reaches a predetermined number of times, it is judged to meet the predetermined weather conditions, and the bad weather level is a foggy level, a rainy level, a snowy level, or a dusty level.

[0007] In some implementations, the processing module further performs image recognition on at least one frame of the driving image to obtain a visibility index. When the processing module determines that the visibility index is less than a first visibility threshold and also less than a second visibility threshold, or when the processing module determines that the visibility index decreases continuously within a predetermined time and each decrease reaches a default threshold, the processing module adjusts the weather state obtained by the deep learning model to a weather state with a worse visibility level, and the first visibility threshold is greater than the second visibility threshold.

[0008] In other embodiments, the driving recorder further includes a global positioning system module electrically connected to the processing module. The processing module determines a road type based on the driving image, and the road type includes a general road type and a highway type. The processing module also calculates a vehicle speed based on the geographic coordinate information obtained by the global positioning system module, and determines the shooting time and uploading frequency corresponding to the predetermined interval time-lapse image according to the road type and the vehicle speed.

[0009] In some embodiments, when the processing module determines that the weather condition meets the predetermined weather condition, the road type is the general road type, and the vehicle speed is greater than a first predetermined speed, the time-lapse image of the predetermined interval is uploaded to the remote server for storage every first predetermined period, and the duration of the time-lapse image of the predetermined interval is equal to the driving image of the first predetermined period, with the frame rate reduced to a first predetermined duration per minute.

[0010] In other embodiments, when the processing module determines that the weather condition meets the predetermined weather condition, and the road type is the highway type, and the vehicle speed is greater than a second predetermined speed and less than a third predetermined speed, the time-lapse image of the predetermined interval is uploaded to the remote server for storage every second predetermined period, and the duration of the time-lapse image of the predetermined interval is equal to the driving image of the second predetermined period, reduced to the second predetermined duration at frames per minute.

[0011] In other embodiments, when the processing module determines that the weather condition meets the predetermined weather condition, and the road type is the expressway type, and the vehicle speed is greater than the third predetermined speed, the time-lapse image of the predetermined interval is uploaded to the remote server for storage every third predetermined period, and the duration of the time-lapse image of the predetermined interval is equal to the driving image of the third predetermined period, reduced in frames per minute to a third predetermined duration.

[0012] Another aspect of the present invention provides an image storage method applicable to a vehicle driving recorder and a remote server, and comprising steps (A) to (C).

[0013] In step (A), a vehicle image is captured by an image capture module of the vehicle recorder.

[0014] In step (B), the driving image is stored in a first storage device of the driving recorder.

[0015] In step (C), a processing module of the driving recorder transmits a predetermined interval of the driving image to the remote server for storage via the communication module when the current weather condition meets the predetermined weather condition.

[0016] In some embodiments, the image storage method further includes a step (D) between steps (B) and (C), wherein the processing module inputs the plurality of frames of the driving image into a deep learning model to obtain the weather condition corresponding to a judgment score. When the processing module determines that the judgment scores obtained by the deep learning model are gradually approaching the judgment score corresponding to a bad weather level for a predetermined number of times, it is determined that the predetermined weather condition is met. The bad weather level is a fog level, a rain level, a snow level, or a dust level.

[0017] In some implementations, in step (D), the processing module further performs image recognition on at least one frame of the driving image to obtain a visibility index. When the processing module determines that the visibility index is less than a first visibility threshold and also less than a second visibility threshold, or when the processing module determines that the visibility index decreases continuously within a predetermined time and each decrease reaches a default threshold, the processing module adjusts the weather state obtained by the deep learning model to a weather state with a worse visibility level, and the first visibility threshold is greater than the second visibility threshold.

[0018] In some embodiments, in step (C), the processing module further determines a road type based on the driving image. The road type includes a general road type and a highway road type. The processing module also calculates a vehicle speed based on geographic coordinate information obtained by a global positioning system module of the driving recorder, and determines the shooting time and upload frequency corresponding to the predetermined interval time-lapse image based on the road type and the vehicle speed.

[0019] In some embodiments, in step (C), when the processing module determines that the weather condition meets the predetermined weather condition, and the road type is the general road type, and the vehicle speed is greater than a first predetermined speed, the time-lapse image of the predetermined interval is uploaded to the remote server for storage every first predetermined period, and the duration of the time-lapse image of the predetermined interval is equal to the driving image of the first predetermined period, with the frame rate reduced per minute to a first predetermined duration.

[0020] In other embodiments, in step (C), when the processing module determines that the weather condition meets the predetermined weather condition, and the road type is the highway type, and the vehicle speed is greater than a second predetermined speed and less than a third predetermined speed, the time-lapse image of the predetermined interval is uploaded to the remote server for storage every second predetermined period, and the duration of the time-lapse image of the predetermined interval is equal to the driving image of the second predetermined period, reduced to the second predetermined duration at a frame rate per minute.

[0021] In other embodiments, in step (C), when the processing module determines that the weather condition meets the predetermined weather condition, and the road type is the expressway type, and the vehicle speed is greater than the third predetermined speed, the time-lapse image of the predetermined interval is uploaded to the remote server for storage every third predetermined period, and the duration of the time-lapse image of the predetermined interval is equal to the driving image of the third predetermined period, reduced in frames per minute to a third predetermined duration.

[0022] Compared with the prior art, the driving recorder and image saving method of the present invention, by means of the processing module, when the current weather conditions meet the predetermined weather conditions, that is, when the possibility of traffic accidents increases due to bad weather, the predetermined interval time-lapse image corresponding to the driving image is uploaded to the remote server for storage according to the current road type. This ensures that even if the vehicle unfortunately encounters an accident, causing the first storage device to be damaged and the driving image to be lost, the image at the moment of the accident can be backed up on the remote server. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Other features and effects of the present invention will be clearly presented in the embodiments with reference to the accompanying drawings, in which:

[0024] Figure 1 is a block diagram illustrating an embodiment of a driving recorder of the present invention; and

[0025] Figure 2 It is a flow chart illustrating a flow chart of the image preservation method of the present invention. DETAILED DESCRIPTION

[0026] Before the present invention is described in detail, it should be noted that similar elements are denoted by the same reference numerals in the following description.

[0027] See also Figure 1 An embodiment of the driving recorder 100 of the present invention is applicable to a remote server 91, a weather information server 92, and a vehicle, and includes a processing module 1, and an image capture module 2 electrically connected to the processing module 1, a communication module 3, a first storage device 4, a global positioning system (GPS) module 6, and an inertial measurement unit (IMU) 7.

[0028] The image capture module 2 is, for example, a camera lens and a corresponding imaging element. The driving recorder 100 is arranged on the vehicle, and after being started, the processing module 1 controls the image capture module 2 to capture a driving image, such as an image in front of the vehicle. The communication module 3 is used to provide a wireless communication function, such as but not limited to 3G, 4G, 5G or Wi-Fi and other wireless communication functions that meet these standards. The first storage device 4 is pre-set by the driving recorder 100 to store the driving image, and is, for example, a memory card (such as an SD card). The second memory device is a non-volatile memory (such as a flash memory) built into the driving recorder 100. After the driving recorder 100 is started, the processing module 1 stores the driving image in the first storage device 4. The inertial measurement module 7 is used to measure an axial angular velocity value and an axial acceleration value related to the current vehicle as a whole.

[0029] In this embodiment, the processing module 1 identifies and determines whether the current weather state A meets the default weather preset condition based on the driving image. When the processing module 1 determines that the weather preset condition is triggered, a time-lapse image of a predetermined interval of the driving image is transmitted to the remote server 91 via the communication module 3 for storage.

[0030] In more detail, the predetermined weather condition is based on any one of the current rain level, snow level, dust level, and fog level as the basis for determining whether the predetermined weather condition is met. That is, when the rain level, snow level, dust level, and fog level respectively reach the pre-set conditions, it indicates that the current driving weather should be in moderate to heavy rain, moderate to heavy snow, dust environment, or fog environment, in other words, bad weather.

[0031] Further explanation, the processing module 1 inputs multiple frames of the driving image into a deep learning model to obtain the weather conditions of the current driving environment. The deep learning model is, for example, ResNet50. ResNet50 is designed to solve the degradation problem that when constructing the network layer of the image recognition deep learning model, the increase in the number of layers will lead to a decrease in judgment accuracy. ResNet50 is designed with a residual learning mechanism to avoid the degradation problem of the high-level model, so as to increase the number of layers of the overall deep learning model and utilize more image features to enhance the accuracy of image recognition. In this case, the deep learning model of ResNet50 is used to train various weather images, including six main types: clear, cloudy, rainy, foggy, snowy, and sandy. Among them, these main types can be further subdivided according to needs, for example but not limited to: rainy days are subdivided into drizzle, moderate rain, and heavy rain; snow is subdivided into light snow, moderate snow, and heavy snow; sandy is subdivided into general and heavy sandy; fog is subdivided into light fog and dense fog, etc. The processing module 1 uses the continuous multiple frames of images (i.e., the driving images) captured by the image capture module 2, for example but not limited to 15 consecutive frames as a group, to perform image recognition on each frame and count the weather status proportion of all frames in the group, and use the weather status with the highest proportion as the weather status represented by the group. At the same time, the processing module 1 continuously collects the results of a certain number of groups as the data basis for the weather status within a period of time, that is, calculates and determines the weather status at the current time through the weather status results of all groups within the period of time. It is worth mentioning that when it is raining in the driving environment, the rain first forms tiny water droplets on the windshield. With different rainfall levels, it will affect the formation of larger water droplets and short-term water flow on the windshield surface. Therefore, the length of formation time and the different characteristics between images are sufficient to identify and determine which of the above-mentioned subdivisions the current rain level belongs to. Finally, the deep learning model will output a judgment score based on the result of the above-mentioned image judgment. The judgment score may be, for example, between 0 and 1, and a corresponding score interval is defined according to the aforementioned weather condition type. When the judgment score falls within the interval, it means that the processing module 1 judges that the weather condition corresponds to the sunny, cloudy, drizzle, moderate rain, heavy rain, fog, snow and dust states corresponding to the interval. The deep learning model described in this paragraph is not limited to ResNet50, and all deep learning models used in the field of image recognition are implementable models.

[0032] For example, the processing module 1 obtains the judgment score corresponding to the current weather state by the deep learning model, which is 0.679, between the interval of 0.65 to 0.8 corresponding to cloudy, and then judges that the current weather state is in accordance with the weather state within the corresponding score interval. When the deep learning model obtains the judgment score corresponding to the current weather state, which is 0.509, between the interval of 0.5 to 0.65 corresponding to moderate rain, then judges that the current weather state is in accordance with the weather state within the corresponding score interval and meets the predetermined weather conditions. In addition, in other embodiments, the processing module 1 can also be changed to; when the judgment scores obtained by the deep learning model for the current image are all closer to the score corresponding to the weather state (i.e., bad weather) and reach a predetermined number of times (such as 3 times) in a row, then judge that the current weather state is in accordance with the predetermined weather conditions. That is, the current weather state is in accordance with the trend towards bad weather development, such as the degree of fog, the degree of rain, the degree of snow, and the degree of dust. For example, three different driving images are input into the deep learning model over time, and the judgment scores obtained are 0.71, 0.69, and 0.675, which is gradually approaching 0.65 for moderate rain, and is judged to meet the predetermined weather conditions. The aforementioned three judgment scores can be used to determine whether they have trend characteristics through a pre-set gap.

[0033] On the other hand, in other embodiments, the processing module 1 further adjusts the weather condition obtained by the deep learning model according to a visibility index as an auxiliary judgment. For example, when the processing module 1 determines that the current weather condition (i.e., the weather condition obtained by the deep learning model) is a weather condition attribute such as dust or fog that is sensitive and easily causes the recognition result to be unstable due to changes in environmental conditions, the processing module is set to 1 to automatically trigger the collection of at least one frame of the current driving image for image recognition to obtain the visibility index. However, it can also be set for various or all weather conditions (i.e., the weather condition obtained by the deep learning model) to be triggered by the processing module 1 to obtain the visibility index accordingly. The grayscale comparison method (ITCcB) of the paper "Evaluation of Experimental Methods for Measuring Atmospheric Visibility Index by Digital Imaging Technology" by Tsai Dai-ling on July 24, 2014, first calculates the average brightness of a target in the image minus the average brightness of the background, and then divides it by the average brightness of the background to obtain an image grayscale. Then, the global dynamic range, the local dynamic range, and the image grayscale are multiplied by three pre-set weight parameters and then summed up to obtain the visibility index. However, this method does not limit the calculation method of the visibility index.

[0034] One aspect of the auxiliary judgment is: when the processing module 1 determines that the visibility index is less than a first visibility threshold (e.g., 800 meters) but greater than a third visibility threshold (e.g., 500 meters), the weather state corresponding to the current time (i.e., the weather state after adjusting the weather state originally obtained by the deep learning model) can be generated in the same manner according to which visibility threshold stage the visibility index is in. For example, the visibility index is less than the first visibility threshold but greater than the third visibility threshold, corresponding to the current general dust or slight fog, and the result is compared with the previous weather state (i.e., the weather state obtained by the deep learning model). If both are general dust or slight fog, it means that the two are consistent, and the previously determined weather state (i.e., the weather state obtained by the deep learning model) is consistent with the previous weather state. In addition, when it is determined that the visibility index is less than the first visibility threshold and is also less than a second visibility threshold (for example, 200 meters), the second visibility threshold corresponds to the visibility of heavy dust or dense fog, which means that the weather state of general dust or light fog previously judged by the processing module 1 (that is, the weather state obtained by the deep learning model) is inconsistent with the weather state of heavy dust or dense fog judged by the current visibility threshold (that is, the weather state after adjusting the weather state originally obtained by the deep learning model), indicating that the previous judgment may have been a light judgment. The processing module 1 can adjust the weather state to heavy dust or dense fog according to the visibility index to meet the weather default condition. Among them, whether the visibility thresholds match general, heavy dust or light, dense fog is to find out the visibility index that the actual visibility can reasonably correspond to under the conditions of general, heavy dust or light fog, dense fog through actual experiments in advance, and use it as a reference for whether the visibility threshold matches. Another aspect of the auxiliary judgment is: when the processing module 1 starts to calculate the visibility index, and as the weather condition judgment is continuously executed, the visibility index decreases at least twice continuously, and the degree of each decrease is greater than a default threshold (for example, 250 meters), it means that the visibility is rapidly deteriorating, and the processing module 1 will adjust the weather condition of heavy dust or dense fog in advance to meet the weather default condition.By combining the visibility index and the auxiliary judgment, when the processing module 1 judges the weather conditions through image recognition, in the case of fog and dust whose degree of difference is relatively easy to be unclear in the image, the probability of misjudgment is high and the delay function is turned on, the change of the visibility index can be referred to to assist in the judgment of the overall weather conditions. Even if the visibility index drops rapidly in a short period of time, the processing module 1 can also start the corresponding function in advance, and transmit the predetermined interval time-lapse image of the driving image to the remote server 91 through the communication module 3 for storage, ensuring that the driving image of the driving process under bad weather can be appropriately saved in advance in response to the risks generated by the current weather conditions. It is particularly noted that the first visibility threshold is greater than the third visibility threshold and greater than the second visibility threshold.

[0035] In addition, in the above-mentioned embodiments, the processing module 1 can also be changed to receive the signal of the global positioning system module 6 to obtain the current location, and transmit the location to the weather information server 92 via the communication module 3 to obtain the weather status. The weather information server 92 is, for example, an open weather forecast database.

[0036] On the other hand, in other embodiments, when a vehicle is traveling on different types of roads (i.e., a road type), different driving speeds will be generated due to different driving environment types, and the speed and driving environment conditions will undoubtedly have different effects on the severity of traffic accidents involving vehicles. For example, in a highway (frame) road environment, vehicles are not restricted by traffic lights, pedestrians, temporarily parked vehicles, etc., and the vehicle itself is more inclined to drive in a straight line. The overall driving environment is relatively simple. Therefore, when the vehicle speed is set to 70, 80 or even 120 kilometers per hour, the degree of damage is more serious. However, when the highway is congested and the vehicle travels at a speed of 30 to 40 kilometers per hour, the driving mode and route caused by the road environment are relatively simple, and the driver is more likely to be ready to brake at any time on the driving route. The collision damage is more likely to be limited to a certain extent. Therefore, for relatively high-speed conditions, a second predetermined speed and a third predetermined speed are used as the judgment thresholds. Mainly, driving on a flat road is faced with a mixed flow of many different types of vehicles, temporarily parked vehicles, unsafe pedestrian interference, complex road design and various roadbed equipment. Considering the driving environment, a first predetermined speed is taken as the judgment threshold in the same relatively high-speed situation. As far as possible, the relative risks caused by different driving environment conditions for vehicles driving on roads with different attributes are distinguished, and in the case of higher risks, more active measures are taken for the driving image data. Therefore, the processing module 1 also performs image recognition on the driving image, and according to the geographic coordinate information obtained by the global positioning system module 6, through the geographic coordinate information and image recognition, when it is determined that a traffic light or a zebra crossing appears in the driving image recognition within a predetermined distance, it is determined that the road type is a general road type, and when it is determined that no traffic light or zebra crossing appears in the driving image within the predetermined distance, it is determined that the road type is a high-speed road type. In addition, the position change and direction change of the vehicle's driving state can be known through the axial angular velocity value and the axial acceleration value measured by the inertial measurement module 7, and then further confirmed whether a movement pattern of the vehicle constructed by the position change and direction change of the vehicle conforms to a predetermined pattern with a height change (such as a slowly climbing elevated bridge), so as to judge the type of road the vehicle is currently on from the movement pattern as a highway type or a general road type.That is to say, the processing module 1 generates a driving motion mode according to the axial angular velocity value and the axial acceleration value detected by the inertial measurement module 7, and when the processing module 1 determines that the driving motion mode meets a predetermined height change pattern, the processing module 1 determines that the road type is the highway type. In addition, the processing module 1 can also make equivalent judgments and distinctions on the road type based on the vehicle coordinates and the distance changes of the points of interest in the map data, the values ​​of the air pressure sensors, the identification of traffic signs and other various vehicle-mounted information. This is a method that can be understood by those with ordinary knowledge in the field, and the judgment of the road type is not the focus of the present invention, so it will not be repeated here.

[0037] The predetermined interval time-lapse image is, for example, a driving image with a time-lapse photography characteristic and a fixed image length that is recorded starting from a trigger time point, wherein the trigger time point is, for example, the time point corresponding to the weather condition that meets the predetermined weather condition, or the time point when the weather condition is judged to meet the predetermined weather condition during the calculation process. In addition, the time-lapse photography characteristic is achieved by reducing the frame rate (FPS) of the image to compress the span of the shooting time, so that the image content of the continuous shooting time is divided into blocks by the set shooting time length and uniformly fixedly output into, for example, a 60-second image file. In addition, the processing module 1 can also individually adjust the shooting time span and upload frequency corresponding to the predetermined interval time-lapse image according to the road type where the vehicle is located and the size of a vehicle speed. Among them, the processing module 1 calculates the vehicle speed based on the geographic coordinate information obtained by the global positioning system module 6.

[0038] For another example, when the processing module 1 determines that the weather condition meets the predetermined weather condition and the road type is the general road type, based on the corresponding conditions, when the vehicle speed is greater than the first predetermined speed (such as 40 kilometers per hour), the time-lapse image of the predetermined interval is uploaded to the remote server 91 for storage every first predetermined period (such as 6 minutes), and the length of time of the time-lapse image of the predetermined interval is equal to the driving image of the first predetermined period, which is reduced to a first predetermined time length (such as 10 seconds) per minute.

[0039] When the processing module 1 determines that the weather condition meets the predetermined weather condition and the road type is the expressway type, according to the corresponding conditions, when the vehicle speed is greater than the second predetermined speed (such as 75 kilometers per hour) and less than the third predetermined speed (such as 90 kilometers per hour), the time-lapse image of the predetermined interval is uploaded to the remote server 91 for storage every second predetermined period (such as 4 minutes), and the length of time of the time-lapse image of the predetermined interval is equal to the driving image of the second predetermined period, which is reduced to a second predetermined time length (such as 15 seconds) per minute.

[0040] When the processing module 1 determines that the weather condition meets the predetermined weather condition and the road type is the highway type, according to the corresponding condition, when the vehicle speed is greater than the third predetermined speed, the time-lapse image of the predetermined interval is uploaded to the remote server 91 for storage every third predetermined period (such as 2 minutes), and the time length of the time-lapse image of the predetermined interval is equal to the driving image of the third predetermined period, which is reduced to a third predetermined time length (such as 30 seconds) per minute, and the third predetermined period is less than the second predetermined period, and the third predetermined time length is greater than the second predetermined time length. That is, the processing module 1 adopts different predetermined cycles and predetermined durations of processing based on the overall risk formed by the road type (i.e., the road type) and the vehicle speed. When on the expressway type, the vehicle travels at a faster speed, and the overall risk is more affected by bad weather than that of the general road type. At the same time, the opportunity to obtain the driving image from the road environment is lower, so the degree of image time-lapse is lower than that of the general road type, and the image of the driving process is fully presented to a greater extent. At the same time, because the faster the vehicle speed, the higher the possibility that the driving recorder cannot be read once a collision occurs, a more frequent upload frequency is adopted than that of the general road type. Similarly, for vehicles traveling near the upper edge of the speed limit of the expressway type, the proportion of time-lapse photography will be further reduced and the frequency of image upload will be increased, so that the upload frequency of the predetermined interval time-lapse image can be dynamically adjusted to ensure that once an accident occurs, even if the device is damaged and cannot be read, the cloud can still have the driving image at the time of the accident as a record, while taking into account the transmission burden and key content when uploading the image content.

[0041] See also Figure 2 , Figure 2 This is an embodiment of the image storage method of the present invention, which is applicable to the driving recorder 100, the remote server 91, the weather information server 92, and the vehicle, and includes steps S1 to S5. The technical contents of steps S1 to S5 are substantially the same as those of the embodiment of the driving recorder 100 described above.

[0042] In summary, the processing module 1 determines the length and frequency of the time-lapse image of the predetermined interval of the driving image according to the current weather conditions when the predetermined weather conditions are met, and then according to the current road type, so as to upload it to the remote server 91 for storage, thereby achieving the purpose of the present invention.

[0043] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims

1. A driving recorder, suitable for a remote server, characterized in that: And contains: An image capture module, used to capture a vehicle image; A communication module, used to provide networking function; a first storage device, used to store the driving image; and A processing module is electrically connected to the image capture module, the communication module, and the first storage device, and transmits a predetermined interval of the driving image through the communication module to the remote server for storage when the current weather state meets the predetermined weather condition.

2. The driving recorder according to claim 1, characterized in that: The processing module inputs multiple frames of the driving image into a deep learning model to obtain the weather condition corresponding to a judgment score. When the processing module determines that the judgment scores obtained continuously by the deep learning model are gradually approaching the judgment score corresponding to a bad weather level and reaches a predetermined number of times, it is judged that the predetermined weather condition is met, and the bad weather level is a foggy level, a rainy level, a snowy level, or a dusty level.

3. The driving recorder according to claim 2, characterized in that: The processing module also performs image recognition on at least one frame of the driving image to obtain a visibility index. When the processing module determines that the visibility index is less than a first visibility threshold and also less than a second visibility threshold, or when the processing module determines that the visibility index continuously decreases within a predetermined time and each time the decrease reaches a default threshold, the processing module adjusts the weather condition obtained by the deep learning model to a weather condition with a worse visibility level, and the first visibility threshold is greater than the second visibility threshold.

4. The driving recorder according to claim 1, characterized in that: It also includes a global positioning system module electrically connected to the processing module. The processing module determines a road type based on the driving image. The road type includes a general road type and a highway type. The processing module also calculates a vehicle speed based on the geographic coordinate information obtained by the global positioning system module, and determines the shooting time and upload frequency corresponding to the time-lapse image of the predetermined interval based on the road type and the vehicle speed.

5. The driving recorder according to claim 4, characterized in that: When the processing module determines that the weather condition meets the predetermined weather condition, the road type is the general road type, and the vehicle speed is greater than a first predetermined speed, the time-lapse image of the predetermined interval is uploaded to the remote server for storage every first predetermined period, and the time length of the time-lapse image of the predetermined interval is equal to the driving image of the first predetermined period, with the frame rate reduced per minute to a first predetermined time length.

6. The driving recorder according to claim 4, characterized in that: When the processing module determines that the weather condition meets the predetermined weather condition, and the road type is the expressway type, and the vehicle speed is greater than a second predetermined speed and less than a third predetermined speed, the time-lapse image of the predetermined interval is uploaded to the remote server for storage every second predetermined period, and the time length of the time-lapse image of the predetermined interval is equal to the driving image of the second predetermined period, which is reduced to the second predetermined time length by frames per minute.

7. The driving recorder according to claim 4, characterized in that: When the processing module determines that the weather condition meets the predetermined weather condition, the road type is the expressway type, and the vehicle speed is greater than the third predetermined speed, the time-lapse image of the predetermined interval is uploaded to the remote server for storage every third predetermined period, and the time length of the time-lapse image of the predetermined interval is equal to the driving image of the third predetermined period, which is reduced to a third predetermined time length per minute by frames.

8. An image storage method, applicable to a vehicle recorder and a remote server, characterized in that: And contains: (A) capturing a vehicle image by an image capture module of the vehicle driving recorder; (B) storing the driving image by a first storage device of the driving recorder; and (C) When the current weather conditions meet the predetermined weather conditions, a processing module of the driving recorder transmits a predetermined interval of the driving image to the remote server via the communication module for storage.

9. The image storage method according to claim 8, characterized in that: The method further includes a step (D) between steps (B) and (C), wherein the processing module inputs multiple frames of the driving image into a deep learning model to obtain the weather condition corresponding to a judgment score. When the processing module determines that the judgment scores continuously obtained by the deep learning model are gradually approaching the judgment score corresponding to a bad weather level and reach a predetermined number of times, it is determined that the predetermined weather condition is met, and the bad weather level is a foggy level, a rainy level, a snowy level, or a dusty level.

10. The image storage method according to claim 9, characterized in that: In step (D), the processing module also performs image recognition on at least one frame of the driving image to obtain a visibility index. When the processing module determines that the visibility index is less than a first visibility threshold and also less than a second visibility threshold, or when the processing module determines that the visibility index continuously decreases within a predetermined time and each time it decreases to a default threshold, the processing module adjusts the weather condition obtained by the deep learning model to a weather condition with a worse visibility level, and the first visibility threshold is greater than the second visibility threshold.

11. The image storage method according to claim 10, characterized in that: In step (C), the processing module also determines a road type based on the driving image, and the road type includes a general road type and a highway type. The processing module also calculates a vehicle speed based on the geographic coordinate information obtained by a global positioning system module of the driving recorder, and determines the shooting time and upload frequency corresponding to the predetermined interval time-lapse image based on the road type and the vehicle speed.

12. The image storage method according to claim 11, characterized in that: In step (C), when the processing module determines that the weather condition meets the predetermined weather condition, the road type is the general road type, and the vehicle speed is greater than a first predetermined speed, the time-lapse image of the predetermined interval is uploaded to the remote server for storage every first predetermined period, and the duration of the time-lapse image of the predetermined interval is equal to the driving image of the first predetermined period, which is reduced to a first predetermined duration at a frame rate per minute.

13. The image storage method according to claim 11, characterized in that: In step (C), when the processing module determines that the weather condition meets the predetermined weather condition, and the road type is the expressway type, and the vehicle speed is greater than a second predetermined speed and less than a third predetermined speed, the time-lapse image of the predetermined interval is uploaded to the remote server for storage every second predetermined period, and the duration of the time-lapse image of the predetermined interval is equal to the driving image of the second predetermined period, which is reduced to the second predetermined duration at a frame rate per minute.

14. The image storage method according to claim 11, characterized in that: In step (C), when the processing module determines that the weather condition meets the predetermined weather condition, and the road type is the expressway type, and the vehicle speed is greater than the third predetermined speed, the time-lapse image of the predetermined interval is uploaded to the remote server for storage every third predetermined period, and the duration of the time-lapse image of the predetermined interval is equal to the driving image of the third predetermined period, which is reduced to a third predetermined duration at a frame rate per minute.

Images