Abstract: An early warning method for anti-collision is provided. In the method, a vehicle-mounted device obtains image information and radar information and determines a main lane line of a road in a driving direction of a vehicle according to the image information. The vehicle-mounted device determines an obstacle located on the main lane line and obtains motion parameters of the vehicle and the obstacle by fusing of the image information and the radar information, and calculates an estimated collision time period from now to the vehicle collides with the obstacle based on the motion parameters, and calculates a braking time period of the vehicle and outputting an early warning in response that the braking time period is less than the estimated collision time period.

Abstract: A method for training a depth estimation model implemented in an electronic device includes obtaining a first image pair from a training data set; inputting the first left image into the depth estimation model, and obtaining a disparity map; adding the first left image and the disparity map, and obtaining a second right image; calculating a mean square error and cosine similarity of pixel values of all corresponding pixels in the first right image and the second right image; calculating mean values of the mean square error and the cosine similarity, and obtaining a first mean value of the mean square error and a second mean value of the cosine similarity; adding the first mean value and the second mean value, and obtaining a loss value of the depth estimation model; and iteratively training the depth estimation model according to the loss value.

Abstract: A method for improving the reconstruction of images obtains an object image and a reference image and extracts a first edge of the object image and a second edge of the reference image based on a predetermined algorithm. A first vector of the plurality of first pixels to the first edge and a second vector of the plurality of second pixels to the second edge are obtained and a determination made as to whether the first vector and the second vector are consistent. A loss between the first vector and the second vector is calculated if the first and second vector are not consistent and a predetermined model is corrected based on the loss, the reference image being reconstructed into the object image based on corrected predetermined model. An image reconstruction device and a non-transitory storage medium are also disclosed.

Abstract: A method for detecting vehicle deviation, an electronic device and a storage medium are provided. In the method, a first foreground image is acquired, a first corrected image is obtained by performing a distortion correction on the first foreground image and an aerial view is obtained. Based on the aerial view, a distribution map of non-zero pixel points is generated. An initial position of a left lane line and an initial position of a right lane line in the aerial view are determined. A first curve corresponding to the left lane line and a second curve corresponding to the right lane line are fitted. It can be determined whether the vehicle deviates from any lane line according to a distance between the vehicle and any lane line. The method can effectively detect lane lines, and improve an accuracy of identifying lane line detection.

Abstract: A lane line recognition method applied to an electronic device is provided. In the method, the electronic device converts a first foreground image into a bird's-eye view (BEV) image and determines the initial positions of the left lane line and the right lane line of the BEV image and fits a first curve and a second curve after a first sliding window slid each time and fits a third curve and a fourth curve after a second sliding window slid each time. The electronic device recognizes the left lane line according to the second curve and the right lane line according to the fourth curve. The method improves accuracy for recognizing lane lines of a road when a vehicle is moving on the road.

Abstract: An obstacle identification method applied to a vehicle-mounted device is provided. The method includes collecting radar information and original image information. Fused image information is obtained by fusing the radar information with the original image information. Obstacle point cloud and non-obstacle point cloud are obtained from the fused image information, and the non-obstacle point cloud is filtered out from the fused image information. Once at least one radar point cloud group is obtained by grouping the obstacle point cloud, an obstacle corresponding to each group of the at least one radar point cloud group can be identified.

Abstract: A method for collision warning implemented in an electronic device includes fusing obtained radar information and image information; recognizing at least one obstacle in a traveling direction of a vehicle according to the fused radar information and image information; determining motion parameters of the at least one obstacle and the vehicle according to the radar information and the image information; and calculating a collision time between the vehicle and the at least one obstacle according to the motion parameters, and issuing a collision warning.

Abstract: A method of detecting moving objects is provided. The method obtains point cloud data set of a target scene and an image of the target scene. The method detects one or more stationary object areas from the image. The method records point cloud data, which corresponds to the stationary object areas in the point cloud data set, to be first point cloud data. The method determines a velocity range of the first point cloud data according to the first point cloud data. The method further determines whether one or more moving objects are present in the target scene according to second point cloud data and the velocity range of the first point cloud data. The second point cloud data is the point cloud data of the point cloud data set excluding the first point cloud data. A related electronic device and a non-transitory storage medium are provided.

Abstract: A method for estimating depth implemented in an electronic device includes obtaining a first image; inputting the first image into a depth estimation model, and obtaining a first depth image; obtaining a depth ratio factor, the depth ratio factor indicating a relationship between a relative depth and a depth of each pixel in the first depth image; and obtaining depth information of the first depth image according to the first depth image and the depth ratio factor.

Abstract: A method for training a depth estimation model implemented in an electronic device includes obtaining a first image pair from a training data set; inputting the first left image into the depth estimation model, and obtaining a disparity map; adding the first left image and the disparity map, and obtaining a second right image; calculating a mean square error and cosine similarity of pixel values of all corresponding pixels in the first right image and the second right image; calculating mean values of the mean square error and the cosine similarity, and obtaining a first mean value of the mean square error and a second mean value of the cosine similarity; adding the first mean value and the second mean value, and obtaining a loss value of the depth estimation model; and iteratively training the depth estimation model according to the loss value.

Abstract: A method for improving the reconstruction of images obtains an object image and a reference image and extracts a first edge of the object image and a second edge of the reference image based on a predetermined algorithm. A first vector of the plurality of first pixels to the first edge and a second vector of the plurality of second pixels to the second edge are obtained and a determination made as to whether the first vector and the second vector are consistent. A loss between the first vector and the second vector is calculated if the first and second vector are not consistent and a predetermined model is corrected based on the loss, the reference image being reconstructed into the object image based on corrected predetermined model. An image reconstruction device and a non-transitory storage medium are also disclosed.

Abstract: An image detection obtains original image. The original image is corrected to obtain a corrected image. Median filtering is performed on the corrected image to obtain a filtered image. A contrast of the filtered image is adjusted to obtain an adjusted image. Bilateral filtering is performed on the adjusted image to obtain an enhanced image. Defects in the enhanced image is detected. The method can detect defects in images accurately and efficiently.

Abstract: An image detection obtains first images with defects. Each of the first images is corrected and divided to obtain first sub-region images. The first sub-region images are processed to obtain processed first sub-region images. The processed first sub-region images are used to train a neural network to obtain a target mode. Second images are obtained. Each of the second images is corrected and divided to obtain second sub-region images. The second sub-region images are processed to obtain processed second sub-region images. The target model is applied to detect each of the processed second sub-region images to obtain a detection result. The method can detect defects in images accurately and efficiently.

Abstract: A processing method using depth detection includes: an image capturing step implemented by using a camera device to capture a depth image of a device under test (DUT) and a surrounding environment around the DUT, and then using a signal processing module to carry out a modifying process so as to transform the depth image into a three-dimensional (3D) image corresponding in shape to the DUT; a planning step implemented by using the signal processing module to define spraying spots on the 3D image and to calculate normal vectors of the 3D image respectively corresponding in position to the spraying spots; and a spraying step implemented by using a spraying device to sequentially spray portions of the DUT respectively corresponding in position to the spraying spots, in which a spraying direction of the spraying device is substantially parallel to the normal vector of the corresponding spraying spot.