Object detection for event cameras

1. A method for producing a textural image from event information generated by an event camera comprising:

accumulating event information from a plurality of events occurring during successive event cycles across a field of view of the event camera, each event indicating an x,y location within said field of view, a polarity for a change of detected light intensity incident at said x,y location and an event cycle at which said event occurred;

in response to selected event cycles, analysing event information for one or more preceding event cycles to identify one or more regions of interest bounding a respective object to be tracked; and

responsive to a threshold event criterion for a region of interest being met, generating a textural image for the region of interest from event information accumulated from within the region of interest.

2. The method of claim 1 comprising responsive to identifying a plurality of different regions of interest within the field of view of the event camera, allocating each region a respective threshold event criterion.

3. The method of claim 1 wherein said threshold event criterion includes either: a threshold number of events; a threshold number of event cycles; or a threshold number of events occurring within a threshold number of event cycles.

4. The method of claim 1 wherein said analysing is performed periodically.

5. The method of claim 1 further comprising:

in response to selected event cycles, generating a background textural image for at least regions outside of said identified one or more regions of interest from event information accumulated for a plurality of preceding event cycles.

6. The method of claim 1 wherein said background textural image is of a lower resolution than said textural image for the one or more regions of interest and wherein said method further comprises: up-sampling said background textural image; and combining said up-sampled textural image with said textural image for the one or more regions of interest to generate a high resolution image for the field of view of the event camera.

7. The method of claim 5 wherein said step of generating said background textural image is performed at a lower frequency than generating a textural image for any of the or each region of interest.

8. The method of claim 1 wherein said analysing comprises:

in response to selected event cycles, generating a background textural image for the field of view of the event camera at a lower resolution than said textural image for the one or more regions of interest from event information accumulated for a plurality of preceding event cycles; and

analysing said background textural image to identify said one or more regions of interest.

9. The method of claim 1 wherein said object to be tracked comprises one or both of: a face or eyes.

10. A system comprising an event camera, a memory for accumulating event information generated by the event camera and a processor configured to process said accumulated event information according to a method for producing a textural image from event information generated by an event camera comprising:

accumulating event information from a plurality of events occurring during successive event cycles across a field of view of the event camera, each event indicating an x,y location within said field of view, a polarity for a change of detected light intensity incident at said x,y location and an event cycle at which said event occurred;

in response to selected event cycles, analysing event information for one or more preceding event cycles to identify one or more regions of interest bounding a respective object to be tracked; and

responsive to a threshold event criterion for a region of interest being met, generating a textural image for the region of interest from event information accumulated from within the region of interest.

11. A system according to claim 10 further comprising a face detector configured to perform said analysing, said face detector comprising a convolutional neural network comprising a plurality of convolutional layers, a first of said layers being configured to receive said accumulated event information, successive layers of said network being configured to down sample said accumulated event information into at least one lower resolution feature map comprising an array of cells, at least one intermediate layer being configured to retain information generated by a corresponding layer in a previous instance of said face detector and further comprising at least one transformation layer configured to transform a feature map provided by a previous convolutional layer generate a respective probability for each down-sampled cell of said feature map comprising an object at a given scale.

12. A system according to claim 11 comprising a plurality of said transformation layers, each configured to generate a probability for down-sampled cells of respective feature maps comprising an object at respective scales.

13. A system according to claim 11 wherein each said at least one transformation layer comprises a YOLO layer.

14. A system according to claim 12 further comprising an upsampling layer between a first transformation layer and a second transformation layer so that said second transformation layer operates at a smaller scale than said first transformation layer.

15. A system according to claim 12 further comprising a filter to select a cell with a highest probability as the region of interest for the object.

16. A driver monitoring systems (DMS) comprising a system comprising an event camera, a memory for accumulating event information generated by the event camera and a processor configured to process said accumulated event information according to a method for producing a textural image from event information generated by an event camera comprising:

accumulating event information from a plurality of events occurring during successive event cycles across a field of view of the event camera, each event indicating an x,y location within said field of view, a polarity for a change of detected light intensity incident at said x,y location and an event cycle at which said event occurred;

in response to selected event cycles, analysing event information for one or more preceding event cycles to identify one or more regions of interest bounding a respective object to be tracked; and

responsive to a threshold event criterion for a region of interest being met, generating a textural image for the region of interest from event information accumulated from within the region of interest.