Object detection
An object detection device has a controller configured to receive classification data and regions for an image. The regions are prioritized and if a region is selected for further prediction, a photographic feature for the selected region is determined and a re-capture of the selected region is made based on the photographic feature by updating an image capturing device. The re-captured region is re-processed and classification data is re-generated. Then, final object detection for selected region(s) is performed and final object detection for not-selected region(s) is performed.
1. An object detection device comprising a controller configured to:
a) receive image data produced by an image capturing device and process the image data, wherein the image data represents an image and wherein processing the image data comprises generating classification data and partitioning the image data into regions;
b) prioritize one or more of the regions based on heuristics that comprise one or more of:
low classification confidence;
classification type;
object(s) detected in an adjacent or close region;
a detected eye gaze;
said one or more of the regions being previously selected; and
object density;
c) determine that a region is selected for further classification processing if the region has a priority exceeding a threshold limit; and when the region is selected for further classification processing:
d) determine a photographic feature for the selected region;
e) update an image capturing device to adjust the photographic feature for the selected region and cause the updated image capturing device to perform a re-capture of the selected region;
f) re-process the re-captured region; and thereby
g) re-generate classification data; and then
h1) perform final object detection for selected region(s) based on the re-generated classification data; and
h2) perform final object detection for not-selected region(s) based on the classification data.
2. The object detection device according to claim 1 , wherein the controller is further configured to perform object detection for the image utilizing an object detection model.
3. The object detection device according to claim 2 , wherein the object detection model is based on a Deep Neural Network model.
4. The object detection device according to claim 1 , wherein the controller is further configured to determine the photographic feature by determining a photographic shortcoming of the region and determining a corresponding camera feature.
5. The object detection device according to claim 1 , further comprising the image capturing device, wherein the controller is further configured to e) cause the re-capture of the selected region based on the photographic feature by recapturing the selected region in a re-captured image.
6. The object detection device according to claim 1 , wherein the controller is further configured to perform final object detection for a region by proceeding with inference and making final predictions.
7. The object detection device according to claim 1 , wherein the controller is further configured to h2) perform final object detection for not-selected region(s) when no more regions are selected for further prediction.
8. The object detection device according to claim 1 , wherein the controller is further configured to h2) perform final object detection for not-selected region(s) while performing any of, a combination of or all of:
d) determining a photographic feature for the selected region;
e) causing a re-capture of the selected region based on the photographic feature by updating the image capturing device;
f) re-processing the re-captured region; and
g) re-generating classification data.
9. The object detection device according to claim 4 , wherein the controller is further configured to determine that the photographic shortcoming and the corresponding camera feature is one or several taken from the group comprising:
low contrast—changing exposure;
under exposed—changing exposure;
over exposed—changing exposure;
not enough detail—zooming in;
out of focus—re-focusing; and
blurry—changing exposure.
10. The object detection device according to claim 1 , wherein the controller is further configured to g) re-generate classification data based on received classification data for that region.
11. The object detection device according to claim 1 , wherein the threshold limit is absolute.
12. The object detection device according to claim 1 , wherein the threshold limit is relative.
13. The object detection device according to claim 1 , wherein causing the updated image capturing device to perform a re-capture of the selected region comprises causing a re-capture of several regions in one re-capture.
14. The object detection device according to claim 1 , wherein the object detection device is a smartphone or a tablet computer.
15. The object detection device according to claim 1 , wherein the object detection device is an optical see-through device.
16. The object detection device according to claim 1 , wherein the object detection device is arranged to be used in image retrieval, industrial use, robotic vision and/or video surveillance.
17. A method for object detection in an object detection device, wherein the method comprises:
a) receiving image data produced by an image capturing device and process the image data, wherein the image data represents an image and wherein processing the image data comprises generating classification data and partitioning the image data into regions;
b) prioritizing one or more of the regions based on heuristics that comprise one or more of:
low classification confidence;
classification type;
object(s) detected in an adjacent or close region;
a detected eye gaze;
said one or more of the regions being previously selected; and
object density;
c) determining that a region is selected for further classification processing if the region has a priority exceeding a threshold limit; and when the region is selected for further classification processing:
d) determining a photographic feature for the selected region;
e) updating an image capturing device to adjust the photographic feature for the selected region and causing the updated image capturing device to perform a re-capture of the selected region;
f) re-processing the re-captured region; and thereby
g) re-generating classification data; and then
h1) performing final object detection for selected region(s) based on the re-generated classification data; and
h2) performing final object detection for not-selected region(s) based on the classification data.
18. A non-transitory computer-readable medium carrying computer instructions that when loaded into and executed by a controller of an object detection device enables the object detection device to implement a method for object detection in an object detection device, wherein the method comprises:
a) receiving image data produced by an image capturing device and process the image data, wherein the image data represents an image and wherein processing the image data comprises generating classification data and partitioning the image data into regions;
b) prioritizing one or more of the regions based on heuristics that comprise one or more of:
low classification confidence;
classification type;
object(s) detected in an adjacent or close region;
a detected eye gaze;
said one or more of the regions being previously selected; and
object density;
c) determining that a region is selected for further classification processing if the region has a priority exceeding a threshold limit; and when the region is selected for further classification processing:
d) determining a photographic feature for the selected region;
e) updating an image capturing device to adjust the photographic feature for the selected region and causing the updated image capturing device to perform a re-capture of the selected region;
f) re-processing the re-captured region; and thereby
g) re-generating classification data; and then
h1) performing final object detection for selected region(s) based on the re-generated classification data; and
h2) performing final object detection for not-selected region(s) based on the classification data.