Apparatus and methods for image segmentation using machine learning processes

A method for image segmentation comprising:

• applying a process to an input image to generate input image feature data characterizing features of the input image, and input image segmentation data characterizing an initial segmentation of the input image;
• obtaining, for a plurality of reference images, reference image feature data characterizing features of the plurality of reference images, and reference image classification data characterizing classifications of the plurality of reference images;
• generating reference image segmentation data based on the reference image feature data, the reference image classification data, and the input image feature data;
• and
• combining values of the input image segmentation data with values of the reference image segmentation data to generate blended image segmentation data, the blended image segmentation data characterizing a final segmentation of the input image.

A non-transitory, machine-readable storage medium storing instructions that, when executed by at least one processor, cause the at least one processor to perform operations that include:

• applying a process to an input image to generate input image feature data characterizing features of the input image, and input image segmentation data characterizing an initial segmentation of the input image;
• obtaining, for a plurality of reference images, reference image feature data characterizing features of the plurality of reference images, and reference image classification data characterizing classifications of the plurality of reference images;
• generating reference image segmentation data based on the reference image feature data, the reference image classification data, and the input image feature data;
• and
• combining values of the input image segmentation data with values of the reference image segmentation data to generate blended image segmentation data, the blended image segmentation data characterizing a final segmentation of the input image.