Solid-state image capturing system, solid-state image capturing device, information processing device, image processing method, and information processing method
Provided is a solid-state image capturing system that includes a solid-state image capturing device and an information processing device. The solid-state image capturing device includes a first DNN processing unit that executes, on image data, a part of a DNN algorithm by a first DNN to generate a first result. The information processing device includes a second DNN processing unit that executes, on the first result acquired from the solid-state image capturing device, remaining of the DNN algorithm by a second DNN to generate a second result.
1 . An image capturing system, comprising:
an imaging device; and
an information processing device, wherein
the imaging device includes:
an imaging sensor configured to output an analog signal, wherein the imaging sensor comprises a plurality of pixels in a two-dimensional arrangement;
a first Deep-Neural-Network (DNN) processing unit configured to:
execute, on data which is based on the analog signal, a first part of a DNN algorithm; and
generate a first result based on the execution of the first part of the DNN algorithm; and
a first transmitting unit configured to output the first result to the information processing device, and
the information processing device includes:
a second DNN processing unit configured to:
execute, on the first result, a second part of the DNN algorithm; and
generate a second result based on the execution of the second part of the DNN algorithm; and
a second transmitting unit configured to output the second result to an external device.
2 . The image capturing system according to claim 1 , wherein the first result includes:
a region of interest (ROI) of an object, and
intermediate layer data.
3 . The image capturing system according to claim 1 , wherein
the imaging device further includes a first storage configured to store at least the first part of the DNN algorithm, and
the information processing device further includes a second storage configured to store at least the second part of the DNN algorithm.
4 . The image capturing system according to claim 1 , wherein
the imaging device further includes a first control unit configured to control the first DNN processing unit, and
the information processing device further includes a second control unit configured to control the second DNN processing unit.
5 . The image capturing system according to claim 4 , wherein
the first control unit is further configured to:
generate first control information that includes information on the first part of the DNN algorithm; and
transmit the generated first control information to the second control unit, and
the second control unit is further configured to:
generate second control information that includes information on the second part of the DNN algorithm; and
transmit the generated second control information to the first control unit.
6 . The image capturing system according to claim 5 , wherein
the first control unit is further configured to control the first DNN processing unit based on the second control information; and
the second control unit is further configured to control the second DNN processing unit based on the first control information.
7 . The image capturing system according to claim 4 , wherein
the first control unit is further configured to transmit, to the second control unit, an execution completion notification of the first DNN processing unit, and
the second control unit is further configured to transmit, to the first control unit, an execution completion notification of the second DNN processing unit.
8 . The image capturing system according to claim 1 , wherein the information processing device further includes at least one of an application processor or a cloud server.
9 . The image capturing system according to claim 1 , wherein the first DNN processing unit is further configured to detect an object in the data, based on the analog signal and the first result.
10 . The image capturing system according to claim 9 , wherein the second DNN processing unit is further configured to recognize the detected object.
11 . An image capturing device, comprising:
an imaging sensor configured to output an analog signal, wherein the imaging sensor comprises a plurality of pixels in a two-dimensional arrangement;
a Deep-Neural-Network (DNN) processing unit configured to:
execute, on data which is based on the analog signal, a first part of a DNN algorithm; and
generate a first result based on the execution of the first part of the DNN algorithm; and
a transmitting unit configured to transmit the first result to an information processing device, wherein the information processing device is associated with execution of a second part of the DNN algorithm; and
a control unit configured to:
receive control information from the information processing device, wherein the control information includes information on the second part of the DNN algorithm; and
control the DNN processing unit based on the control information.
12 . The image capturing device according to claim 11 , wherein the first result includes intermediate layer data.
13 . The image capturing device according to claim 11 , further comprising a storage configured to store at least the first part of the DNN algorithm.
14 . An information processing device, comprising:
a Deep-Neural-Network (DNN) processing unit configured to:
receive, from an image capturing device, a first result associated with execution of a first part of a DNN algorithm on data, wherein the data is based on an analog signal;
execute, on the first result, a second part of the DNN algorithm; and
generate a second result based on the execution of the second part of the DNN algorithm; and
a control unit configured to:
receive control information from the image capturing device, wherein the control information includes information on the first part of the DNN algorithm; and
control the DNN processing unit based on the control information.
15 . The information processing device according to claim 14 , further comprising a storage configured to store at least the second part of the DNN algorithm.
16 . An image processing method, comprising:
in an image capturing device:
outputting, by an imaging sensor, an analog signal, wherein the imaging sensor comprises a plurality of pixels in a two-dimensional arrangement;
receiving control information from an information processing device;
controlling a Deep-Neural-Network (DNN) processing unit based on the control information;
executing, on data which is based on the analog signal, a first part of a DNN algorithm, wherein the execution of the first part of the DNN algorithm is based on the control of the DNN processing unit;
generating a first result based on the execution of the first part of the DNN algorithm; and
transmitting the first result to the information processing device, wherein
the information processing device is associated with execution of a second part of the DNN algorithm, and
the control information includes information on the second part of the DNN algorithm.
17 . An information processing method, comprising:
in an information processing device:
receiving, from an image capturing device, a first result and control information, wherein
the first result is associated with execution of a first part of a Deep-Neural-Network (DNN) algorithm on data, and
the data is based on an analog signal;
controlling a DNN processing unit based on the control information;
executing, on the first result, a second part of the DNN algorithm based on the control of the DNN processing unit; and
generating a second result based on the execution of the second part of the DNN algorithm, wherein
the image capturing device is associated with the execution of the first part of the DNN algorithm, and
the control information includes information on the first part of the DNN algorithm.
18 . A non-transitory computer-readable medium having stored thereon, computer-executable instructions which, when executed by a computer, cause the computer to execute operations, the operations comprising:
obtaining, from an imaging sensor, an analog signal, wherein the imaging sensor comprises a plurality of pixels in a two-dimensional arrangement,
receiving control information from an information processing device;
controlling a Deep-Neural-Network (DNN) processing unit based on the control information;
executing, on data which is based on the analog signal, a first part of a DNN algorithm, wherein the execution of the first part of the DNN algorithm is based on the control of the DNN processing unit;
generating a first result based on the execution of the first part of the DNN algorithm; and
transmitting the first result to the information processing device, wherein
the information processing device is associated with execution of a second part of the DNN algorithm, and
the control information includes information on the second part of the DNN algorithm.
19 . A non-transitory computer-readable medium having stored thereon, computer-executable instructions which, when executed by a computer, cause the computer to execute operations, the operations comprising:
receiving, from an image capturing device, a first result and control information, wherein
the first result is associated with execution of a first part of a Deep-Neural-Network (DNN) algorithm on data, and
the data is based on an analog signal;
controlling a DNN processing unit based on the control information;
executing, on the first result, a second part of the DNN algorithm based on the control of the DNN processing unit; and
generating a second result based on the execution of the second part of the DNN algorithm, wherein
the image capturing device is associated with the execution of the first part of the DNN algorithm, and
the control information includes information on the first part of the DNN algorithm.