Content generating device, content distribution system, content generating method, and content generating program
According to one or more embodiments, a content generating device is provided. The content generating device comprises a first generator, a second generator, and a first transmitter. The first generator generates low-bit-rate encoded data that is original data having been encoded to a low-bit-rate. The second generator generates machine-learned model data for generating improved data obtained by improving human perceptually the low-bit-rate encoded data, based on a machine-learned model. The first transmitter transmits the low-bit-rate encoded data and the model data to outside.
1 . A content generating device, comprising at least one processor configured to:
obtain original data including category information;
encode the original data to a low-bit-rate and generate low-bit-rate encoded data;
generate, based on a machine-learned model, machine-learned model data corresponding to the category information of the original data by using teacher data corresponding to the category information, the machine-learned model data to be used for generating improved data that is human perceptually improved from the low-bit-rate encoded data; and
transmit the low-bit-rate encoded data and the model data to outside the device.
2 . The content generating device of claim 1 , wherein the at least one processor is further configured to:
store the model data generated in the past; and
transmit, to the outside of the device, the model data generated in the past together with the low-bit-rate encoded data.
3 . The content generating device of claim 2 , wherein
the original data includes the category information for classifying content according to characteristics thereof, and
the at least one processor records the model data classified based on the category information or an initial value of the model data.
4 . The content generating device of claim 1 , wherein
the original data is image data, and
the low-bit-rate encoded data includes meta-information.
5 . The content generating device of claim 4 , wherein the meta-information of the low-bit-rate encoded data is at least one of: a coding block quantization parameter (QP), a prediction error coefficient, prediction mode information, and motion vector information in an image coding technique.
6 . A content distribution system, comprising:
a content generating device including at least one processor configured to:
obtain original data including category information;
encode the original data to a low-bit-rate and generate low-bit-rate encoded data;
generate, based on a machine-learned model, machine-learned model data corresponding to the category information of the original data by using teacher data corresponding to the category information, the machine-learned model data to be used for generating improved data that is human perceptually improved from the low-bit-rate encoded data; and
transmit the low-bit-rate encoded data and the model data to outside the content generating device; and
a content distribution device including at least one processor configured to:
receive the low-bit-rate encoded data and the model data transmitted from the content generating device;
generate the improved data based on the low-bit-rate encoded data, from the low-bit-rate encoded data and the model data received; and
distribute the improved data as content data.
7 . The content distribution system of claim 6 , wherein the at least one processor of the content generating device is further configured to:
store the model data generated in the past; and
transmit, to the outside the content generating device, the model data generated in the past together with the low-bit-rate encoded data.
8 . The content distribution system of claim 7 , wherein
the original data includes the category information for classifying content according to characteristics thereof, and
the at least one processor of the content generating device records the model data classified based on the category information or an initial value of the model data.
9 . The content distribution system of claim 6 , wherein
the original data is image data, and
the low-bit-rate encoded data includes meta-information.
10 . The content distribution system of claim 9 ,
wherein the meta-information of the low-bit-rate encoded data is at least one of: a coding block quantization parameter (QP), a prediction error coefficient, prediction mode information, and motion vector information in an image coding technique.
11 . A content generating method, comprising:
obtaining original data including category information;
encoding the original data to a low-bito rate and generating low-bit-rate encoded data;
generating, based on a machine-learned model, machine-learned model data corresponding to the category information of the original data by using teacher data corresponding to the category information, the machine-learned model data to be used for generating improved data that is human perceptually improved from the low-bit-rate encoded data; and
transmitting the low-bit-rate encoded data and the model data to outside a device.
12 . The content generating method of claim 11 , further comprising:
receiving the low-bit-rate encoded data and the model data; and
generating the improved data based on the low-bit-rate encoded data, from the low-bit-rate encoded data and the model data received.
13 . The content generating method of claim 11 , further comprising:
storing the model data generated in the past; and
transmitting the model data generated in the past together with the low-bit-rate encoded data to the outside the device.
14 . The content generating method of claim 13 , wherein
the original data includes the category information for classifying content according to characteristics thereof, and
the method further comprises recording the model data classified based on the category information or an initial value of the model data.
15 . The content generating method of claim 11 , wherein
the original data is image data, and
the low-bit-rate encoded data includes meta-information.
16 . The content generating method of claim 15 , wherein the meta-information of the low-bit-rate encoded data is at least one of: a coding block quantization parameter (QP), a prediction error coefficient, prediction mode information, and motion vector information in an image coding technique.
17 . A non-transitory computer-readable medium storing a program that, when executed, causes a computer to execute the content generating method of claim 11 .
18 . The non-transitory computer-readable medium of claim 17 , wherein the method further comprises:
receiving the low-bit-rate encoded data and the model data; and
generating the improved data based on the low-bit-rate encoded data, from the low-bit-rate encoded data and the model data received.
19 . The non-transitory computer-readable medium of claim 17 , wherein the method further comprises:
storing the model data generated in the past; and
transmitting the model data generated in the past together with the low-bit-rate encoded data to the outside the device.
20 . The non-transitory computer-readable medium of claim 17 , wherein
the original data is image data, and
the low-bit-rate encoded data includes meta-information.