Method and apparatus with neural network operation

A processing apparatus comprising:

• a processor;
• and
• a memory storing instructions, wherein execution of the instructions by the processor configures the processor to:
• generate a first intermediate operation result by performing a first-order partial operation of a neural network layer on a first input line of a first area of a frame, generate a second intermediate operation result by performing another first-order partial operation of the neural network layer on a second input line of the first area, and generate an objective operation result of the neural network layer with respect to the first area based on a second-order partial operation performed on the first intermediate operation result and the second intermediate operation result.

An electronic device comprising:

• a camera configured to generate an image frame;
• and
• a processor configured to:
• generate a first intermediate operation result by performing a first-order partial operation of a neural network layer on a first input line of a first area of a frame;
• generate a second intermediate operation result by performing another first-order partial operation of the neural network layer on a second input line of the first area;
• and
• generate an objective operation result of the neural network layer with respect to the first area based on a second-order partial operation performed on the first intermediate operation result and the second intermediate operation result.

A processing apparatus comprising:

• a processor configured to:
• control an incremental receipt of image information of a frame in input line units;
• generate a plurality of output lines of a neural network layer of a neural network by, for each of the plurality of output lines of the neural network layer, accumulating results of one-dimensional (1D) convolution operations on each of respective plural line units of image information of the incrementally received image information of the frame;
• and
• generate a plurality of output lines of a subsequent neural network layer of the neural network by, for each of the plurality of output lines of the subsequent neural network layer, accumulating results of 1D convolution or pooling operations on each of respective multiple output lines of the plurality of output lines of the neural network layer,
• wherein generation of the plurality of output lines of the subsequent neural network layer are performed before all image information of the frame are incrementally received in the incremental receipt of the image information of the frame in the input line units.