Object detection at different illumination conditions
A method for image processing at different illumination conditions, the method may include acquiring an image of an environment of a vehicle; selecting a set of pixels located within a region of interest that is located at an upper part of the image; calculating an illumination condition indicator based on values of the set of pixels; selecting a selected machine learning process, out of a machine learning processes, based on the illumination condition indicator; wherein different machine learning processes are trained to different illumination conditions; and processing the image by the selected machine learning process to provide processing results.
1. A method for image processing at different illumination conditions, the method comprises:
acquiring an image of an environment of a vehicle;
selecting a set of pixels located within a region of interest that is located at an upper part of the image;
calculating an illumination condition indicator based on values of the set of pixels;
selecting a selected machine learning process, out of a plurality of machine learning processes, based on the illumination condition indicator;
wherein different machine learning processes are trained to different illumination conditions;
processing the image by the selected machine learning process to provide processing results; and
dynamically updating a location of pixels and a number of pixels of the set of pixels based on the processing results.
2. The method according to claim 1 comprising dynamically updating a distribution of pixels of the set of pixels within the region of interest based on the processing results.
3. The method according to claim 1 comprising dynamically updating a number of pixels of the set of pixels based on the processing results.
4. The method according to claim 1 comprising evaluating a quality of the processing; and determining, based on the quality of the processing, whether to amend at least one of a location of pixels of the set of pixels and a number of pixels of the set of pixels.
5. The method according to claim 1 comprising receiving or generating a rough estimate of a location of a sky and determining the region of interest to fall within the rough estimate of the location of the sky.
6. The method according to claim 1 comprising deactivating non-selected machine learning processes of the plurality of the machine learning processes.
7. The method according to claim 1 wherein the selecting of the set of pixels, the calculating of the illumination condition indicator and the selecting of the selected machine learning process are executed in real time.
8. The method according to claim 1 wherein the machine learning processes are a night machine learning process and a day machine learning process.
9. The method according to claim 1 wherein the set of pixels are a grid of spaced apart pixels.
10. The method according to claim 1 comprising:
estimating a location of a sky in a certain image; and
estimating a location of the sky in a future image based on (a) the estimated location of the sky in the certain image and (b) an estimated change in a field of view of a camera that acquired the certain image between a time of acquisition of the certain image and an estimated change in a field of view at the future image.
11. The method according to claim 1 comprising determining a feature of pixels of the set of pixels based on available resources for performing a calculating of the different illumination conditions.
12. The method according to claim 1 wherein the processing comprises lane detection.
13. The method according to claim 1 comprising defining a region of interest of one frame based on a location of object bounding boxes in a previous frame.
14. The method according to claim 1 comprising calculating illumination condition indicator values for different shapes within boundaries defined by estimations of vanishing point and objects locations; and selecting a region of interest which has a lowest condition illumination indicator as the region of interest that is located at the upper part of the image.
15. The method according to claim 1 wherein the processing comprises performing at least partial autonomous driving.
16. A non-transitory computer readable medium that stores instructions for:
acquiring an image of an environment of a vehicle;
selecting a set of pixels located within a region of interest that is located at an upper part of the image;
calculating an illumination condition indicator based on values of the set of pixels;
selecting a selected machine learning process, out of a plurality of machine learning processes, based on the illumination condition indicator;
wherein different machine learning processes are trained to different illumination conditions;
processing the image by the selected machine learning process to provide processing results; and
dynamically updating a location of pixels and a number of pixels of the set of pixels based on the processing results.
17. A device comprising at least one unit that is configured to acquire an image of an environment of a vehicle;
select a set of pixels located within a region of interest that is located at an upper part of the image;
calculate an illumination condition indicator based on values of the set of pixels;
select a selected machine learning process, out of a plurality of machine learning processes, based on the illumination condition indicator;
wherein different machine learning processes are trained to different illumination conditions;
process the image by the selected machine learning process to provide processing results; and
dynamically updating a location of pixels and a number of pixels of the set of pixels based on the processing results.