Detecting a cropped image based on an object segmentation mask

A system, comprising:

• a memory;
• and
• at least one processor coupled to the memory and configured to:
• generate a standardized mask from an image segmentation mask depicting an object within an image by:
• identifying a minimum bounding box configured to surround the image segmentation mask within the image;
• removing a portion of the image outside the bounding box from the image;
• and
• resizing the image based on a predefined number of rows of pixels in the image and a predefined number of columns of pixels in the image;
• modify the image by adding a predefined number of pixels between an edge of the image and the standardized mask;
• generate a mask edge by:
• generating an eroded version of the standardized mask using an erosion kernel;
• and
• removing the eroded version of the standardized mask from the standardized mask;
• determine the image is cropped by:
• generating a first vector by, for each row within the number of rows of pixels in the image:
• generating a summation of pixel values along the row;
• and
• storing the summation within the first vector at an index corresponding to a position of the row within the number of rows;
• generating a second vector by, for each column within the number of columns of pixels in the image:
• generating a summation of pixel values along the column;
• and
• storing the summation within the second vector at an index corresponding to a position of the column within the number of columns;
• and
• identifying a summation within the first vector or second vector that is greater than a predefined threshold.

A non-transitory computer-readable device having instructions stored thereon that, when executed by at least one computing device, cause the at least one computing device to perform operations comprising:

• generating a standardized mask from an image segmentation mask depicting an object within an image by:
• identifying a minimum bounding box configured to surround the image segmentation mask within the image;
• removing a portion of the image outside the bounding box from the image;
• and
• resizing the image based on a predefined number of rows of pixels in the image and a predefined number of columns of pixels in the image;
• modifying the image by adding a predefined number of pixels between an edge of the image and the standardized mask;
• generating a mask edge by:
• generating an eroded version of the standardized mask using an erosion kernel;
• and
• removing the eroded version of the standardized mask from the standardized mask;
• determining the image is cropped by:
• generating a first vector by, for each row within the number of rows of pixels in the image:
• generating a summation of pixel values along the row;
• and
• storing the summation within the first vector at an index corresponding to a position of the row within the number of rows;
• generating a second vector by, for each column within the number of columns of pixels in the image:
• generating a summation of pixel values along the column;
• and
• storing the summation within the second vector at an index corresponding to a position of the column within the number of columns;
• and
• identifying a summation within the first vector or second vector that is greater than a predefined threshold.