System and method for detecting and redacting inappropriate images

1 . A system comprising:

one or more processors; and

one or more non-transitory computer-readable media storing computing instructions that, when run on the one or more processors, cause the one or more processors to perform:

capturing, via a camera, an image for upload, wherein the image features a user body of a user;

determining, by the one or more processors, a polygon mask layer having one or more polygon masks for the user body in the image, comprising:

determining, via a body pose landmarking algorithm, key joint points for the user body in the image based on joint landmarks defined in a configuration set; and

determining the one or more polygon masks based on the key joint points and one or more regions of interest defined in the configuration set;

determining, by the one or more processors, a skin mask layer with a skin mask for the image, wherein the skin mask is determined based on a respective color of pixels of the image;

blending the one or more polygon masks and the skin mask into a final mask layer comprising a final mask for the image; and

applying the final mask to the image.

2 . The system in claim 1 , wherein the computing instructions, when run on the one or more processors, further cause the one or more processors to perform:

before determining the polygon mask layer, one or more of:

training the body pose landmarking algorithm to determine body joint landmarks for a human body based on body poses in a first training image dataset and respective landmarks for each of the body poses;

verifying, via the body pose landmarking algorithm, as trained, that a user body pose for the user body in the image corresponds to a predetermined pose in the configuration set; or

receiving, via a computer network, the configuration set from a remote database.

3 . The system in claim 1 , wherein the joint landmarks defined in the configuration set comprise upper body joint landmarks and lower body joint landmarks.

4 . The system in claim 1 , wherein:

the configuration set further comprises one or more geometric formulas configured to define the one or more regions of interest; and

the one or more geometric formulas are associated with the joint landmarks and one or more reference points.

5 . The system in claim 4 , wherein:

the one or more geometric formulas include one or more of:

one or more line segment formulas associated with one or more first pairs of the joint landmarks and the one or more reference points; or

one or more curve segment formulas associated with one or more second pairs of the joint landmarks and the one or more reference points.

6 . The system in claim 1 , wherein determining the skin mask layer further comprises:

determining, via a skin segmentation algorithm, a respective binary value for each pixel of the image based on the respective color for the each pixel.

7 . The system in claim 6 , wherein the computing instructions, when run on the one or more processors, further cause the one or more processors to perform:

before determining the skin mask layer, training the skin segmentation algorithm to determine whether an image pixel is categorized as a skin pixel based on human-skin pixels and non-human-skin pixels of training images in a second training image dataset.

8 . The system in claim 1 , wherein:

blending the one or more polygon masks and the skin mask into the final mask layer comprises performing a bitwise operation on the polygon mask layer and the skin mask layer.

9 . The system in claim 1 , wherein applying the final mask to the image further comprises:

overlaying the final mask layer onto the image to cover a portion of the user body in the image by the final mask.

10 . The system in claim 9 , wherein overlaying the final mask layer onto the image further comprises:

applying a mask style to the final mask; and

overlaying the final mask layer, with the mask style applied to the final mask, onto the image.

11 . A method being implemented via execution of computing instructions configured to run at one or more processors and stored at one or more non-transitory computer-readable media, the method comprising: capturing, via a camera, an image for upload, wherein the image features a user body of a user;

determining, by the one or more processors, a polygon mask layer having one or more polygon masks for the user body in the image, comprising:

determining, via a body pose landmarking algorithm, key joint points for the user body in the image based on joint landmarks defined in a configuration set; and

determining the one or more polygon masks based on the key joint points and one or more regions of interest defined in the configuration set;

determining, by the one or more processors, a skin mask layer with a skin mask for the image, wherein the skin mask is determined based on a respective color of pixels of the image;

blending the one or more polygon masks and the skin mask into a final mask layer comprising a final mask for the image; and

applying the final mask to the image.

12 . The method in claim 11 further comprising:

before determining the polygon mask layer, one or more of:

training the body pose landmarking algorithm to determine body joint landmarks for a human body based on body poses in a first training image dataset and respective landmarks for each of the body poses;

verifying, via the body pose landmarking algorithm, as trained, that a user body pose for the user body in the image corresponds to a predetermined pose in the configuration set; or

receiving, via a computer network, the configuration set from a remote database.

13 . The method in claim 11 , wherein the joint landmarks defined in the configuration set comprise upper body joint landmarks and lower body joint landmarks.

14 . The method in claim 11 , wherein:

the configuration set further comprises one or more geometric formulas configured to define the one or more regions of interest; and

the one or more geometric formulas are associated with the joint landmarks and one or more reference points.

15 . The method in claim 14 , wherein:

the one or more geometric formulas include one or more of:

one or more line segment formulas associated with one or more first pairs of the joint landmarks and the one or more reference points; or

one or more curve segment formulas associated with one or more second pairs of the joint landmarks and the one or more reference points.

16 . The method in claim 11 , wherein determining the skin mask layer further comprises:

determining, via a skin segmentation algorithm, a respective binary value for each pixel of the image based on the respective color for the each pixel.

17 . The method in claim 16 further comprising:

before determining the skin mask layer, training the skin segmentation algorithm to determine whether an image pixel is categorized as a skin pixel based on human-skin pixels and non-human-skin pixels of training images in a second training image dataset.

18 . The method in claim 11 , wherein:

blending the one or more polygon masks and the skin mask into the final mask layer comprises performing a bitwise operation on the polygon mask layer and the skin mask layer.

19 . The method in claim 11 , wherein applying the final mask to the image further comprises:

overlaying the final mask layer onto the image to cover a portion of the user body in the image by the final mask.

20 . The method in claim 19 , wherein overlaying the final mask layer onto the image further comprises:

applying a mask style to the final mask; and

overlaying the final mask layer, with the mask style applied to the final mask, onto the image.