Method for identifying moving vehicles and in-vehicle device
A method for identifying road vehicles or other objects which are in motion against those which are not moving applied in an in-vehicle device of an assisted vehicle which is being driven shoots a first image of a target vehicle and a second later image of the target vehicle, determines a first mask area of the target vehicle from the first image, and determines a second mask of the target vehicle from the second image based on an instance segmentation algorithm. An Intersection over Union (IoU) is calculated between the first mask area and the second mask area and a determination made as to whether a dynamic class object mask area of the target vehicle according to the IoU should be generated. A dynamic class object mask area of the target vehicle is generated when the target vehicle is found to be a moving vehicle.
1 . A method for identifying moving vehicles comprising:
capturing a first image of a target vehicle at a first time and a second image of the target vehicle at a second time;
determining a first mask area of the target vehicle from the first image, and determining a second mask of the target vehicle from the second image based on an instance segmentation algorithm, wherein the first mask area is an obscured area of the first image determined by obscuring a first object in the first image, and the second mask area is an obscured area of the second image determined by obscuring a second object in the second image;
calculating an Intersection over Union (IoU) between the first mask area and the second mask area;
determining whether to generate a dynamic class object mask area of the target vehicle according to the IoU, wherein the dynamic class object mask area is generated by connecting the first mask area and the second mask area, and the dynamic class object mask area is an indication that the vehicle is moving; and
determining that the target vehicle is a moving vehicle if generating the dynamic class object mask area of the target vehicle is determined.
2 . The method for identifying moving vehicles as claimed in claim 1 , further comprising:
determining a first location of the target vehicle in the first image and an identification of the target vehicle.
3 . The method for identifying moving vehicles as claimed in claim 2 , further comprising:
determining a second location of the target vehicle in the second image based on the identification of the target vehicle.
4 . The method for identifying moving vehicles as claimed in claim 3 , further comprising:
inputting the first image into a feature extraction network, and obtaining a first feature map of the first image;
inputting the second image into the feature extraction network, and obtaining a second feature map of the second image;
performing binary classification and coordinate regression on the first feature map and the second feature map, and determining first Regions of Interest (ROIs) of the target vehicle in the first feature map and second ROIs of the target vehicle in the second feature map;
extracting first feature submaps from the first ROIs, and extracting second feature submaps from the first ROIs;
generating the first mask area based on the first feature submaps and the first location;
generating the second mask area based on the second feature submaps and the second location.
5 . The method for identifying moving vehicles as claimed in claim 1 , further comprising:
calculating a first coincidence degree between the first mask area and the second mask area, wherein the first coincidence degree is an intersection;
calculating a second coincidence degree between the first mask area and the second mask area, wherein the second coincidence degree is a union;
calculating a ratio between the first coincidence degree and the second coincidence degree, and defining the ratio as the IoU.
6 . The method for identifying moving vehicles as claimed in claim 1 , further comprising:
comparing the IoU between the first mask area and the second mask area with a preset threshold;
when the IoU is greater than or equal to the preset threshold, generating the dynamic class object mask area of the target vehicle; and
when the IoU between the first mask area and the second mask area is less than the preset threshold, not generating the dynamic class object mask area of the target vehicle.
7 . The method for identifying moving vehicles as claimed in claim 1 , further comprising:
establishing a vehicle safety distance model, and determining a safe braking distance between vehicles based on the vehicle safety distance model;
when determining that a distance between the moving vehicle and one vehicle is less than or equal to the safe braking distance, issuing an alarm message.
8 . The method for identifying moving vehicles as claimed in claim 7 , further comprising:
when the distance between the moving vehicle and the vehicle is equal to the safe braking distance, issuing a first alarm message at a first frequency;
when the distance between the moving vehicle and the vehicle is less than the safe braking distance, issuing the second alarm message at a second frequency, wherein the first frequency is less than the second frequency.
9 . An in-vehicle device comprising:
a processor; and
a non-transitory storage medium coupled to the processor and configured to store a plurality of instructions, which cause the processor to:
capture a first image of a target vehicle at a first time and a second image of the target vehicle at a second time;
determine a first mask area of the target vehicle from the first image, and determine a second mask of the target vehicle from the second image based on an instance segmentation algorithm, wherein the first mask area is an obscured area of the first image determined by obscuring a first object in the first image, and the second mask area is an obscured area of the second image determined by obscuring a second object in the second image;
calculate an Intersection over Union (IoU) between the first mask area and the second mask area;
determine whether to generate a dynamic class object mask area of the target vehicle according to the IoU, wherein the dynamic class object mask area is generated by connecting the first mask area and the second mask area, and the dynamic class object mask area is an indication that the vehicle is moving;
determine that the target vehicle is a moving vehicle if generating the dynamic class object mask area of the target vehicle is determined.
10 . The in-vehicle device as claimed in claim 9 , wherein the plurality of instructions are further configured to cause the processor to:
determine a first location of the target vehicle in the first image and an identification of the target vehicle.
11 . The in-vehicle device as claimed in claim 10 , wherein the plurality of instructions are further configured to cause the processor to:
determine a second location of the target vehicle in the second image based on the identification of the target vehicle.
12 . The in-vehicle device as claimed in claim 11 , wherein the plurality of instructions are further configured to cause the processor to:
input the first image into a feature extraction network, and obtain a first feature map of the first image;
input the second image into the feature extraction network, and obtain a second feature map of the second image;
perform binary classification and coordinate regression on the first feature map and the second feature map, and determine first Regions of Interest (ROIs) of the target vehicle in the first feature map and second ROIs of the target vehicle in the second feature map;
extract first feature submaps from the first ROIs, and extract second feature submaps from the first ROIs;
generate the first mask area based on the first feature submaps and the first location;
generate the second mask area based on the second feature submaps and the second location.
13 . The in-vehicle device as claimed in claim 9 , wherein the plurality of instructions are further configured to cause the processor to:
calculate a first coincidence degree between the first mask area and the second mask area, wherein the first coincidence degree is an intersection;
calculate a second coincidence degree between the first mask area and the second mask area, wherein the second coincidence degree is a union;
calculate a ratio between the first coincidence degree and the second coincidence degree, and define the ratio as the IoU.
14 . The in-vehicle device as claimed in claim 9 , wherein the plurality of instructions are further configured to cause the processor to:
compare the IoU between the first mask area and the second mask area with a preset threshold;
when the IoU is greater than or equal to the preset threshold, generate the dynamic class object mask area of the target vehicle;
when the IoU between the first mask area and the second mask area is less than the preset threshold, fail to generate the dynamic class object mask area of the target vehicle.
15 . The in-vehicle device as claimed in claim 9 , wherein the plurality of instructions are further configured to cause the processor to:
establish a vehicle safety distance model, and determining a safe braking distance between vehicles based on the vehicle safety distance model;
when determining that a distance between the moving vehicle and one vehicle is less than or equal to the safe braking distance, issue an alarm message.
16 . The in-vehicle device as claimed in claim 15 , wherein the plurality of instructions are further configured to cause the processor to:
when the distance between the moving vehicle and the vehicle is equal to the safe braking distance, issue a first alarm message at a first frequency;
when the distance between the moving vehicle and the vehicle is less than the safe braking distance, issue the second alarm message at a second frequency, wherein the first frequency is less than the second frequency.
17 . A non-transitory storage medium having stored thereon instructions that, when executed by at least one processor of an in-vehicle device, causes the least one processor to execute instructions of a method for identifying moving vehicles, the method comprising:
capturing a first image of a target vehicle at a first time and a second image of the target vehicle at a second time;
determining a first mask area of the target vehicle from the first image, and determining a second mask of the target vehicle from the second image based on an instance segmentation algorithm, wherein the first mask area is an obscured area of the first image determined by obscuring a first object in the first image, and the second mask area is an obscured area of the second image determined by obscuring a second object in the second image;
calculating an Intersection over Union (IoU) between the first mask area and the second mask area;
determining whether to generate a dynamic class object mask area of the target vehicle according to the IoU, wherein the dynamic class object mask area is generated by connecting the first mask area and the second mask area, and the dynamic class object mask area is an indication that the vehicle is moving;
determining that the target vehicle is a moving vehicle of generating the dynamic class object mask area of the target vehicle is determined.
18 . The non-transitory storage medium as recited in claim 17 , wherein the method comprising:
determining a first location of the target vehicle in the first image and an identification of the target vehicle.
19 . The non-transitory storage medium as recited in claim 18 , wherein the method comprising:
determining a second location of the target vehicle in the second image based on the identification of the target vehicle.
20 . The non-transitory storage medium as recited in claim 19 , wherein the method comprising:
inputting the first image into a feature extraction network, and obtaining a first feature map of the first image;
inputting the second image into the feature extraction network, and obtaining a second feature map of the second image;
performing binary classification and coordinate regression on the first feature map and the second feature map, and determining first Regions of Interest (ROIs) of the target vehicle in the first feature map and second ROIs of the target vehicle in the second feature map;
extracting first feature submaps from the first ROIs, and extracting second feature submaps from the first ROIs;
generating the first mask area based on the first feature submaps and the first location;
generating the second mask area based on the second feature submaps and the second location.