Road detecting method and apparatus

1. A road detection method, comprising:

acquiring an image of a predetermined region;

semantically segmenting the image to acquire a first probability of a region corresponding to each pixel in the image being a road region;

acquiring a historical position information set of a target terminal;

correcting, in response to historical position information of the target terminal existing in the historical position information set, the historical position information of the target terminal indicating a historical position of the target terminal located in the predetermined region, the first probability according to the historical position information of the target terminal to obtain a second probability, wherein the historical position information of the target terminal includes coordinate information uploaded by the target terminal within a predetermined period; and

determining a region corresponding to a pixel having the second probability greater than a preset threshold as a road region.

2. The method according to claim 1 , wherein the semantically segmenting the image to acquire a first probability of a region corresponding to each pixel in the image being a road region comprises:

importing the image into a pre-trained fully convolutional network model to obtain a third probability of a region corresponding to the each pixel in the image being a road region; and

determining the first probability according to the third probability and a pre-trained conditional random field model with the each pixel in the image as a node in a conditional random field and a relation between pixels in the image as an edge in the conditional random field.

3. The method according to claim 1 , wherein

the correcting the first probability according to the historical position information to obtain a second probability comprises:

counting a quantity of coordinate points as indicated by the coordinate information included in a region corresponding to the each pixel in the image;

determining a fourth probability of the region corresponding to the each pixel in the image being a road region according to the quantity; and

weighting the first probability and the fourth probability of the each pixel according to a preset weight to obtain the second probability of the each pixel.

4. The method according to claim 3 , wherein the counting a quantity of coordinate points as indicated by the coordinate information included in a region corresponding to the each pixel in the image comprises:

establishing a distribution map of the coordinate points according to the coordinate information;

performing image filtering and denoising on the distribution map to obtain a non-noise coordinate point in the distribution map; and

counting a quantity of non-noise coordinate points included in the region corresponding to the each pixel in the image.

5. The method according to claim 1 , wherein the acquiring a historical position information set of a target terminal comprises:

acquiring a position information set uploaded when a preset process and/or a preset thread in the target terminal are started, wherein the preset process and the preset thread are used for realizing a navigation function.

6. A road detection apparatus, comprising:

at least one processor; and

a memory storing instructions, which when executed by the at least one processor, cause the at least one processor to perform operations, the operations comprising:

acquiring an image of a predetermined region;

semantically segmenting the image to acquire a first probability of a region corresponding to each pixel in the image being a road region;

acquiring a historical position information set of a target terminal;

correcting, in response to historical position information of the target terminal existing in the historical position information set, the historical position information of the target terminal indicating a historical position of the target terminal located in the predetermined region, the first probability according to the historical position information of the target terminal to obtain a second probability, wherein the historical position information of the target terminal includes coordinate information uploaded by the target terminal within a predetermined period; and

determining a region corresponding to a pixel having the second probability greater than a preset threshold as a road region.

7. The apparatus according to claim 6 , wherein the semantically segmenting the image to acquire a first probability of a region corresponding to each pixel in the image being a road region comprises:

importing the image into a pre-trained fully convolutional network model to obtain a third probability of a region corresponding to the each pixel in the image being a road region; and

determining the first probability according to the third probability and a pre-trained conditional random field model with the each pixel in the image as a node in a conditional random field and a relation between pixels in the image as an edge in the conditional random field.

8. The apparatus according to claim 6 , wherein

the correcting the first probability according to the historical position information to obtain a second probability comprises:

counting a quantity of coordinate points as indicated by the coordinate information included in a region corresponding to the each pixel in the image;

determining a fourth probability of the region corresponding to the each pixel in the image being a road region according to the quantity; and

weighting the first probability and the fourth probability of the each pixel according to a preset weight to obtain the second probability of the each pixel.

9. The apparatus according to claim 8 , wherein the counting a quantity of coordinate points as indicated by the coordinate information included in a region corresponding to the each pixel in the image comprises:

establishing a distribution map of the coordinate points according to the coordinate information;

performing image filtering and denoising on the distribution map to obtain a non-noise coordinate point in the distribution map; and

counting a quantity of non-noise coordinate points included in the region corresponding to the each pixel in the image.

10. The apparatus according to claim 6 , wherein the acquiring a historical position information set of a target terminal comprises:

acquiring a position information set uploaded when a preset process and/or a preset thread in the target terminal are started, wherein the preset process and the preset thread are used for realizing a navigation function.

11. A non-transitory computer readable storage medium storing a computer program, wherein the program, when executed by a processor, causes the processor to perform the method according to claim 1 .