PARTIAL FACE DETECTION FOR VIDEO CONFERENCING

Abstract

In one aspect, a device may include at least one processor and storage accessible to the at least one processor. The storage may include instructions executable by the at least one processor to provide at least a first image to a model as input, where the model may have been trained using groups of training images that show respective different parts of faces of one or more people but not full faces of the one or more people. The instructions may also be executable to receive, as output from the model, an indication regarding a facial body part. Based on the indication, the instructions may be executable to present a notification during a video conference, where the notification may indicate at least one action to take for a full face to be shown in a video stream as part of the video conference.

Description

Claims

1. A device, comprising: at least one processor; andstorage accessible to the at least one processor and comprising instructions executable by the at least one processor to: during deployment of a model, provide at least a first image to the model as input to the model, the model trained using groups of training images that show respective different parts of faces of one or more people but not full faces of the one or more people;receive, as output from the model, an indication of a facial body part shown in the first image; andbased on the indication, present a notification to a participant of a video conference, the notification indicating at least one action for the participant to take for the participant’s full face to be shown in a video stream as part of the video conference.

2. The device of claim 1, wherein the instructions are executable to: train the model using the groups of training images that show respective different parts of faces of the one or more people but not full faces of the one or more people.

3. The device of claim 2, wherein the training images are labeled with respective facial body parts shown in the respective training images.

4. The device of claim 2, wherein the instructions are executable to: train the model using at least one group of training images that show ceilings but no facial body parts.

5. The device of claim 2, wherein the instructions are executable to: train the model using at least one group of training images that show keyboards but no facial body parts.

6. The device of claim 2, wherein the instructions are executable to: train the model using at least one group of training images that show one or more of: desk tops but no facial body parts, table tops but no facial body parts.

7. The device of claim 1, wherein the notification indicates an adjustment that the participant is to make to a camera for the participant’s full face to be shown in the video stream as part of the video conference.

8. The device of claim 7, comprising the camera.

9. The device of claim 1, wherein the notification indicates a body posture adjustment that the participant is to make for the participant’s full face to be shown in the video stream as part of the video conference.

10. The device of claim 1, wherein the notification indicates a body positioning adjustment that the participant is to make for the participant’s full face to be shown in the video stream as part of the video conference.

11. The device of claim 1, wherein the model comprises one or more artificial neural networks.

12. A method, comprising: providing at least a first image to a model as input, the model trained using groups of training images that show respective different parts of faces of one or more people but not full faces of the one or more people;receiving, as output from the model, an indication regarding a facial body part; andbased on the indication, presenting a notification to a participant of a video conference, the notification indicating at least one action for the participant to take for the participant’s full face to be shown in a video stream as part of the video conference.

13. The method of claim 12, wherein the indication indicates that no facial body part is shown in the first image.

14. The method of claim 12, wherein the indication indicates that a full face is not shown in the first image.

15. The method of claim 14, wherein the indication still indicates at least one facial body part shown in the first image.

16. The method of claim 12, comprising: training the model using the groups of training images that show respective different parts of faces of the one or more people but not full faces of the one or more people.

17. The method of claim 16, wherein the training images are labeled with respective facial body parts shown in the respective training images.

18. The method of claim 12, wherein the notification indicates an adjustment that the participant is to make to a camera for the participant’s full face to be shown in the video stream as part of the video conference.

19. At least one computer readable storage medium (CRSM) that is not a transitory signal, the at least one computer readable storage medium comprising instructions executable by at least one processor to: provide at least a first image to a model as input, the model trained using groups of training images that show respective different parts of faces of one or more people but not full faces of the one or more people;receive, as output from the model, an indication regarding a facial body part; andbased on the indication, present a notification during a video conference, the notification indicating at least one action to take for a full face to be shown in a video stream as part of the video conference.

20. The at least one CRSM of claim 19, wherein the instructions are executable to: identify first and second participants of the video conference, the first and second participants both using a same device to participate in the video conference; andpresent the notification, the notification indicating at least one action to take for full faces of both the first and second participants to be shown in the video stream as part of the video conference.