Semantic image segmentation using gated dense pyramid blocks

An apparatus for semantic image segmentation, comprising:

• an image receiver to receive an image to be segmented;
• a gated dense pyramid network comprising a plurality of gated dense pyramid (GDP) blocks to be trained to generate semantic labels for each pixel in the received image;
• and
• an image generator to generate a segmented image based on the generated semantic labels.

A method for training gated dense pyramid networks, comprising:

• forward training, via a processor, a GDP block comprising a dense pyramid network;
• backward training, via the processor, the dense pyramid network with a gated function;
• and
• co-training, via the processor, the gated dense pyramid network on a whole of the GDP block.

At least one computer readable medium for segmenting images semantically using gated dense pyramid networks having instructions stored therein that, in response to being executed on a computing device, cause the computing device to:

• train a gated dense pyramid (GDP) block of a GDP network to generate semantic labels;
• receive an image to be segmented;
• process the image, via the GDP network, to generate a semantic label for each pixel in the image;
• and
• generate a segmented image based on the received semantic labels.