Provided are a device and operating method thereof for obtaining compression ratio information for recognizing a target object in an image using a deep neural network model, and compressing an image using the compression ratio information and encoding the compressed image. According to an embodiment of the present disclosure, there is provided a device that receives an image via at least one camera or a communication interface, obtains a feature map for detecting a target object in the received image, outputs a compression ratio for correctly recognizing the target object in the image by inputting the image and the feature map to a deep neural network model composed of pre-trained model parameters, and generates a bitstream by compressing the image using the output compression ratio and encoding the compressed image.

A load balancing method for video decoding. The load balancing includes first determining which hardware devices are suitable for the new decoding process, and determining the current load of each of the suitable hardware devices. From the suitable devices potential devices are selected having a current load less than a threshold and overloaded devices are selected having a load greater than or equal to the threshold. If there are no suitable devices, then the decoding process is implemented by software decoding. If the list of potential hardware devices includes only one potential hardware device, then the decoding process is implemented on the hardware device. If the list of potential hardware devices includes more than one potential hardware device, then it is determined how many decoding processes are currently running on each potential hardware device, and the new decoding process is implemented on the potential hardware device having the fewest processes.

A load balancing method for video decoding. The load balancing includes first determining which hardware devices are suitable for the new decoding process, and determining the current load of each of the suitable hardware devices. From the suitable devices potential devices are selected having a current load less than a threshold and overloaded devices are selected having a load greater than or equal to the threshold. If there are no suitable devices, then the decoding process is implemented by software decoding. If the list of potential hardware devices includes only one potential hardware device, then the decoding process is implemented on the hardware device. If the list of potential hardware devices includes more than one potential hardware device, then it is determined how many decoding processes are currently running on each potential hardware device, and the new decoding process is implemented on the potential hardware device having the fewest processes.

Video decoding innovations for multithreading implementations and graphics processor unit (“GPU”) implementations are described. For example, for multithreaded decoding, a decoder uses innovations in the areas of layered data structures, picture extent discovery, a picture command queue, and/or task scheduling for multithreading. Or, for a GPU implementation, a decoder uses innovations in the areas of inverse transforms, inverse quantization, fractional interpolation, intra prediction using waves, loop filtering using waves, memory usage and/or performance-adaptive loop filtering. Innovations are also described in the areas of error handling and recovery, determination of neighbor availability for operations such as context modeling and intra prediction, CABAC decoding, computation of collocated information for direct mode macroblocks in B slices, reduction of memory consumption, implementation of trick play modes, and picture dropping for quality adjustment.

Video decoding innovations for multithreading implementations and graphics processor unit (“GPU”) implementations are described. For example, for multithreaded decoding, a decoder uses innovations in the areas of layered data structures, picture extent discovery, a picture command queue, and/or task scheduling for multithreading. Or, for a GPU implementation, a decoder uses innovations in the areas of inverse transforms, inverse quantization, fractional interpolation, intra prediction using waves, loop filtering using waves, memory usage and/or performance-adaptive loop filtering. Innovations are also described in the areas of error handling and recovery, determination of neighbor availability for operations such as context modeling and intra prediction, CABAC decoding, computation of collocated information for direct mode macroblocks in B slices, reduction of memory consumption, implementation of trick play modes, and picture dropping for quality adjustment.

The present technology relates to an image processing device and an image processing method which allow a deblocking filtering process to apply filtering appropriately. A pixel (p0.sub.i) of which the value is 255 (solid line) before a deblocking process changes greatly to 159 (dot line) after a conventional deblocking process. Therefore, a clipping process having a clipping value of 10 is performed in strong filtering, whereby the pixel (p0.sub.i) of which the value is 255 (solid line) before the deblocking process becomes 245 (bold line). Thus, a change in the pixel value occurring in the conventional technique can be suppressed as much as possible. This disclosure can be applied to an image processing device, for example.

The present technology relates to an image processing device and an image processing method which allow a deblocking filtering process to apply filtering appropriately. A pixel (p0.sub.i) of which the value is 255 (solid line) before a deblocking process changes greatly to 159 (dot line) after a conventional deblocking process. Therefore, a clipping process having a clipping value of 10 is performed in strong filtering, whereby the pixel (p0.sub.i) of which the value is 255 (solid line) before the deblocking process becomes 245 (bold line). Thus, a change in the pixel value occurring in the conventional technique can be suppressed as much as possible. This disclosure can be applied to an image processing device, for example.

In one embodiment, the disclosure provides a computer-implemented method for Progressive Subsampled Transmission of image data. In one embodiment, a source computer may: generate a first down-sampled frame by sampling an input image according to a first sampling pattern; generate a first encoded down-sampled frame; transmit the first encoded down-sampled frame to a recipient device to cause the recipient device to display/use a first output frame generated by decoding and up-sampling the first encoded down-sampled frame; generate a second down-sampled frame by sampling the input image according to a second sampling pattern; generate a second encoded down-sampled frame; and transmit the second encoded down-sampled frame to the recipient device to cause the recipient device to display/use a second output frame generated based on the first encoded down-sampled frame and the second encoded down-sampled frame and in accordance with the first sampling pattern and the second sampling pattern.

In one embodiment, the disclosure provides a computer-implemented method for Progressive Subsampled Transmission of image data. In one embodiment, a source computer may: generate a first down-sampled frame by sampling an input image according to a first sampling pattern; generate a first encoded down-sampled frame; transmit the first encoded down-sampled frame to a recipient device to cause the recipient device to display/use a first output frame generated by decoding and up-sampling the first encoded down-sampled frame; generate a second down-sampled frame by sampling the input image according to a second sampling pattern; generate a second encoded down-sampled frame; and transmit the second encoded down-sampled frame to the recipient device to cause the recipient device to display/use a second output frame generated based on the first encoded down-sampled frame and the second encoded down-sampled frame and in accordance with the first sampling pattern and the second sampling pattern.

A method of converting 10-bit pixel data (e.g. 10:10:10:2 data) into 8-bit pixel data involves converting the 10-bit values to 7-bits or 8-bits and generating error values for each of the converted values. Two of the 8-bit output channels comprise a combination of a converted 7-bit value and one of the bits from the fourth input channel. A third 8-bit output channel comprises the converted 8-bit value and the fourth 8-bit output channel comprises the error values. In various examples, the bits of the error values may be interleaved when they are packed into the fourth output channel.