Image processing apparatus and image processing method
An image processing apparatus comprises a first tracking circuit configured to apply machine learning (ML)-based first tracking processing to images and a second tracking circuit configured to apply non-ML-based second tracking processing to the images. The apparatus controls operations of the first and second tracking circuits so that an operation frequency of the first tracking circuit to be lower than an operation frequency of the second tracking circuit.
1 . An image processing apparatus comprising:
one or more processors that execute a program stored in a memory which cause the image processing apparatus to:
apply machine learning (ML)-based first tracking processing to images, wherein a result of the first tracking processing used is autofocus processing; and
apply non-ML-based second tracking processing to the images,
wherein an operation frequency of the first tracking processing to be lower than an operation frequency of the second tracking processing by deactivating the first tracking processing for an image to which the autofocus processing is not performed.
2 . The image processing apparatus according to claim 1 , wherein the images are for display, and a result of the second tracking processing are used for live-view display.
3 . The image processing apparatus according to claim 2 , wherein the result of the second tracking processing is used to superimpose an image to indicate a tracked subject on the live-view display.
4 . The image processing apparatus according to claim 2 , wherein if a result of the first tracking processing is available, it is used for the live-view display.
5 . The image processing apparatus according to claim 1 , wherein the images are frames of a moving image, the second tracking processing is applied every frame, and the first tracking processing is applied once every plurality of frames.
6 . The image processing apparatus according to claim 5 , wherein the second tracking processing is not applied to a frame for which a result of the first tracking processing is available.
7 . The image processing apparatus according to claim 5 , wherein the number of the plurality of frames when the image processing apparatus satisfies a predetermined condition is more than the number of the plurality of frames when the image processing apparatus does not satisfy the predetermined condition.
8 . The image processing apparatus according to claim 7 , wherein the predetermined condition is that the image processing apparatus is set to an operation mode for reducing power consumption.
9 . The image processing apparatus according to claim 1 , wherein the autofocus processing uses a focus detection region set in accordance with a result of the first tracking processing.
10 . The image processing apparatus according to claim 1 , wherein the machine learning is deep learning.
11 . The image processing apparatus according to claim 1 , wherein the second tracking processing is based on pattern matching.
12 . The image processing apparatus according to claim 1 , wherein a time required to perform the first tracking processing to an image is not greater than an operation cycle of the autofocus processing.
13 . An image processing apparatus comprising:
one or more processors that execute a program stored in a memory and thereby causes the image processing apparatus to:
acquire images having a predetermined frame rate;
apply machine learning (ML)-based first tracking processing to the images;
apply non-ML-based second tracking processing to the images; and
execute processing in which a result of the first tracking processing or the second tracking processing is used,
wherein a time required for the first tracking processing is longer than one frame period of the images, and a time required for the second tracking processing is shorter than the one frame period, and
if results of the first and second tracking processing are available at a start of a process performed for every frame, the result of the first tracking processing rather than the result of the second tracking processing is used even if the result of the first tracking process is for an older frame than a frame for which the result of the second tracking processing for a previous frame is available.
14 . An image capture apparatus comprising:
an image sensor; and
an image processing apparatus using images obtained by the image sensor,
wherein the image processing apparatus comprising:
one or more processors that execute a program stored in a memory and thereby causes the image processing apparatus to:
apply machine learning (ML)-based first tracking processing to images, wherein a result of the first tracking processing used is autofocus processing;
apply non-ML-based second tracking processing to the images; and
control operations of the first and second tracking units,
wherein an operation frequency of the first tracking processing to be lower than an operation frequency of the second tracking processing by deactivating the first tracking processing for an image to which the autofocus processing is not performed.
15 . An image capture apparatus comprising:
an image sensor; and
an image processing apparatus using an image obtained by the image sensor,
wherein the image processing apparatus comprising:
one or more processors that execute a program stored in a memory and thereby causes the image processing apparatus to:
acquire images having a predetermined frame rate;
apply machine learning (ML)-based first tracking processing to the images;
apply non-ML-based second tracking processing to the images; and
execute processing in which a result of the first tracking processing or the second tracking processing is used,
wherein a time required for the first tracking processing is longer than one frame period of the images, and a time required for the second tracking processing is shorter than the one frame period, and
if results of the first and second tracking processing are available at a start of a process performed for every frame, the result of the first tracking processing rather than the result of the second tracking processing are used even if the result of the first tracking process is for an older frame than a frame for which the result of the second tracking processing for a previous frame is available.
16 . An image processing method to be executed by an image processing apparatus, the image processing apparatus including a first tracking circuit configured to apply machine learning (ML)-based first tracking processing to images and a second tracking circuit configured to apply non-ML-based second tracking processing to the images, wherein a result of the first tracking processing used is autofocus processing and the image processing method comprising:
controlling operations of the first and second tracking circuits so that an operation frequency of the first tracking circuit to be lower than an operation frequency of the second tracking circuit by deactivating the first tracking circuit for an image to which the autofocus processing is not performed.
17 . A non-transitory machine-readable medium that has stored therein a program for causing a computer to function as an image processing apparatus comprising the operations of:
applying machine learning (ML)-based first tracking processing to images, wherein a result of the first tracking processing used is autofocus processing;
applying non-ML-based second tracking processing to the images; and
controlling operations of the first and second tracking units,
wherein an operation frequency of the first tracking processing to be lower than an operation frequency of the second tracking processing by deactivating the first tracking processing for an image to which the autofocus processing is not performed.
18 . A non-transitory machine-readable medium that has stored therein a program for causing a computer to function as an image processing apparatus comprising the operations of:
acquiring images having a predetermined frame rate;
applying machine learning (ML)-based first tracking processing to the images;
applying non-ML-based second tracking processing to the images; and
executing processing in which a result of the first tracking processing or the second tracking processing is used,
wherein a time required for the first tracking processing is longer than one frame period of the images, and a time required for the second tracking processing is shorter than the one frame period, and
if results of the first and second tracking processing are available at a start of a process performed for every frame, the result of the first tracking processing rather than the result of the second tracking processing is used even if the result of the first tracking process is for an older frame than a frame for which the result of the second tracking processing for a previous frame is available.