Real-time image-based vehicle detection based on a multi-stage classification
The present disclosure is directed to an autonomous vehicle having a vehicle control system. The vehicle control system includes a vehicle detection system. The vehicle detection system includes receiving an image of a field of view of the vehicle and identifying a region-pair in the image with a sliding-window filter. The region-pair is made up of a first region and a second region. Each region is determined based on a color of pixels within the sliding-window filter. The vehicle detection system also determines a potential second vehicle in the image based on the region-pair. In response to determining the potential second vehicle in the image, the vehicle detection system performs a multi-stage classification of the image to determine whether the second vehicle is present in the image. Additionally, the vehicle detection system provides instructions to control the first vehicle based at least on the determined second vehicle.
1. A method comprising:
receiving, from an image-capture device coupled to a first vehicle, an image of a field of view of the first vehicle;
identifying at least one region-pair in the image with a sliding-window filter, wherein the region-pair comprises a first region of the image and a second region of the image and wherein there is a space between the first region of the image and the second region of the image;
determining a potential second vehicle in the image based on an edge detection of the space between the first region of the image and the second region of the image;
in response to determining a presence of the potential second vehicle in the image, performing a multi-stage classification of the image as a whole to determine whether the second vehicle is present in the image, wherein the multi-stage classification is ended based on a given stage resulting in a determination of absence of the second vehicle in the image; and
providing instructions to control, using a computing device, the first vehicle based at least on the second vehicle being determined to be present in the image.
2. The method of claim 1 , wherein the sliding-window filter defines a plurality of regions based on the color of pixels within the sliding-window filter.
3. The method of claim 1 , wherein each region of the image is determined based on a color of pixels within the sliding-window filter.
4. The method of claim 3 , wherein determining a potential second vehicle further comprises comparing a difference in the color of pixels between the first region and the second region.
5. The method of claim 1 , wherein a size of the sliding-window filter is based on a location on the image.
6. The method of claim 1 , wherein multi-stage classification of the image comprises performing a plurality of classifications of the image, wherein each subsequent classification performs a more advanced and computationally intensive vehicle detection.
7. The method of claim 1 , wherein providing instructions to control comprises providing instructions to perform one or more of: reducing a speed of the first vehicle, maintaining a predetermined safe distance from the second vehicle, avoiding being in a blind spot of the second vehicle, changing lanes to move away from the second vehicle, and stopping the first vehicle.
8. A control system comprising:
at least one processor; and
a memory having stored thereon instructions that, upon execution by the at least one processor, cause the control system to perform functions comprising:
receiving, from an image-capture device coupled to a first vehicle, an image of a field of view of the first vehicle;
identifying at least one region-pair in the image with a sliding-window filter, wherein the region-pair comprises a first region of the image and a second region of the image and wherein there is a space between the first region of the image and the second region of the image;
determining a potential second vehicle in the image based on an edge detection of the space between the first region of the image and the second region of the image;
in response to determining a presence of the potential second vehicle in the image, performing a multi-stage classification of the image as a whole to determine whether the second vehicle is present in the image, wherein the multi-stage classification is ended based on a given stage resulting in a determination of absence of the second vehicle in the image; and
providing instructions to control, using a computing device, the first vehicle based at least on the second vehicle being determined to be present in the image.
9. The control system of claim 8 , wherein the sliding-window filter defines a plurality of regions based on the color of pixels within the sliding-window filter.
10. The control system of claim 8 , wherein each region of the image is determined based on a color of pixels within the sliding-window filter.
11. The control system of claim 10 , wherein determining a potential second vehicle further comprises comparing a difference in the color of pixels between the first region and the second region.
12. The control system of claim 8 , wherein a size of the sliding-window filter is based on a location on the image.
13. The control system of claim 8 , wherein classifying the image comprises performing a plurality of classifications of the image, wherein each subsequent classification performs a more advanced and computationally intensive vehicle detection.
14. A non-transitory computer readable medium having stored thereon instructions, that when executed by a computing device, cause the computing device to perform functions comprising:
receiving, from an image-capture device coupled to a vehicle, an image of a field of view of the first vehicle;
identifying at least one region-pair in the image with a sliding-window filter, wherein the region-pair comprises a first region of the image and a second region of the image and wherein there is a space between the first region of the image and the second region of the image;
determining a potential second vehicle in the image based on an edge detection of the space between the first region of the image and the second region of the image;
in response to determining a presence of the potential second vehicle in the image, performing a multi-stage classification of the image as a whole to determine whether the second vehicle is present in the image, wherein the multi-stage classification is ended based on a given stage resulting in a determination of absence of the second vehicle in the image; and
providing instructions to control the vehicle based at least on the determined second vehicle.
15. The non-transitory computer readable medium of claim 14 , wherein the sliding-window filter defines a plurality of regions based on the color of pixels within the sliding-window filter.
16. The non-transitory computer readable medium of claim 14 , wherein each region of the image is determined based on a color of pixels within the sliding-window filter.
17. The non-transitory computer readable medium of claim 16 , wherein determining a potential second vehicle further comprises comparing a difference in the color of pixels between the first region and the second region.
18. The non-transitory computer readable medium of claim 14 , wherein the size of the sliding-window filter operates as a function of a location on the image.
19. The non-transitory computer readable medium of claim 14 , wherein multi-stage classification of the image comprises performing a plurality of classifications of the image, wherein each subsequent classification performs a more advanced and computationally intensive vehicle detection.
20. The non-transitory computer readable medium of claim 14 , wherein providing instructions to control comprises providing instructions to perform one or more of: reducing a speed of the first vehicle, maintaining a predetermined safe distance from the second vehicle, avoiding being in a blind spot of the second vehicle, changing lanes to move away from the second vehicle, and stopping the first vehicle.