Object detection with modified image background
An embodiment of a semiconductor package apparatus may include technology to pre-process an image to simplify a background of the image, and perform object detection on the pre-processed image with the simplified background. For example, an embodiment of a semiconductor package may include technology to pre-process an image to subtract the background from the image and perform object detection on the pre-processed image with the background subtracted. Other embodiments are disclosed and claimed.
1. An electronic processing system, comprising:
a processor;
memory communicatively coupled to the processor; and
circuitry communicatively coupled to the processor to:
process an image to modify a background of the image to a uniform color;
perform object detection on the processed image with the modified background; and
detect an object in the processed image with modified background based on a convolutional neural network.
2. The system of claim 1 , further comprising:
an image processor to modify the background of the image to provide the processed image.
3. The system of claim 1 , wherein the circuitry is further to:
process the image to split the image into a foreground image with the modified background and the background image;
perform object detection on both the foreground image and the background image; and
combine results of the object detection on the foreground image with results of the object detection on the background image.
4. The system of claim 3 , wherein the circuitry is further to:
determine object occlusion for the background image based on two or more image frames; and
fill in the background image based on the determined object occlusion.
5. The system of claim 3 , wherein the circuitry is further to:
perform object detection on the foreground image with a first model; and
perform object detection on the background image with a second model which is different from the first model.
6. The system of claim 1 , wherein the circuitry is further to:
perform the object detection on the processed image with a single shot detector model.
7. A semiconductor package apparatus, comprising:
one or more substrates; and
circuitry coupled to the one or more substrates, wherein the circuitry is at least partly implemented in one or more of configurable circuitry and fixed-functionality hardware circuitry, the circuitry coupled to the one or more substrates to:
process an image to modify a background of the image to a uniform color;
perform object detection on the processed image with the modified background; and
detect an object in the processed image with modified background based on a convolutional neural network.
8. The apparatus of claim 7 , wherein the circuitry comprises:
an image processor to modify the background of the image to provide the pre processed image.
9. The apparatus of claim 7 , wherein the circuitry is further to:
process the image to split the image into a foreground image with the modified background and the background image;
perform object detection on both the foreground image and the background image; and
combine results of the object detection on the foreground image with results of the object detection on the background image.
10. The apparatus of claim 9 , wherein the circuitry is further to:
determine object occlusion for the background image based on two or more image frames; and
fill in the background image based on the determined object occlusion.
11. The apparatus of claim 9 , wherein the circuitry is further to:
perform object detection on the foreground image with a first model; and
perform object detection on the background image with a second model which is different from the first model.
12. The apparatus of claim 7 , wherein the circuitry is further to:
perform the object detection on the processed image with a single shot detector model.
13. The apparatus of claim 7 , wherein the circuitry coupled to the one or more substrates includes transistor channel regions that are positioned within the one or more substrates.
14. A method comprising:
processing an image to modify a background of the image to a uniform color; and
performing object detection on the processed image with the modified background; and
detecting an object in the processed image with modified background based on a convolutional neural network.
15. The method of claim 14 , further comprising:
processing the image to split the image into a foreground image with the modified background and a background image;
performing object detection on both the foreground image and the background image; and
combining results of the object detection on the foreground image with results of the object detection on the background image.
16. The method of claim 15 , further comprising:
determining object occlusion for the background image based on two or more image frames; and
filling in the background image based on the determined object occlusion.
17. The method of claim 15 , further comprising:
performing object detection on the foreground image with a first model; and
performing object detection on the background image with a second model which is different from the first model.
18. The method of claim 14 , further comprising:
performing the object detection on the processed image with a single shot detector model.
19. At least one non-transitory computer readable medium, comprising a set of instructions, which when executed by a computing device, cause the computing device to:
process an image to modify a background of the image to a uniform color;
perform object detection on the processed image with the modified background; and
detect an object in the processed image with modified background based on a convolutional neural network.
20. The at least one computer readable medium of claim 19 , comprising a further set of instructions, which when executed by the computing device, cause the computing device to:
process the image to split the image into a foreground image with the modified background and the background image;
perform object detection on both the foreground image and the background image; and
combine results of the object detection on the foreground image with results of the object detection on the background image.
21. The at least one computer readable medium of claim 20 , comprising a further set of instructions, which when executed by the computing device, cause the computing device to:
determine object occlusion for the background image based on two or more image frames; and
fill in the background image based on the determined object occlusion.
22. The at least one computer readable medium of claim 20 , comprising a further set of instructions, which when executed by the computing device, cause the computing device to:
perform object detection on the foreground image with a first model; and
perform object detection on the background image with a second model which is different from the first model.
23. The at least one computer readable medium of claim 19 , comprising a further set of instructions, which when executed by the computing device, cause the computing device to:
perform the object detection on the processed image with a single shot detector model.