Image recognition device, solid-state imaging device, and image recognition method
Provided are an image recognition device, a solid-state imaging device, and an image recognition method capable of improving accuracy in recognizing a subject. The image recognition device according to the present disclosure includes an imaging unit and a recognition unit. The imaging unit captures a plurality of images having different sensitivities in one frame period to generate image data of the plurality of images. The recognition unit recognizes the subject from the image data of each of the images, and recognizes the subject captured in an image of one frame based on a result of recognizing the subject.
1. An image recognition device, comprising:
an imaging unit configured to:
capture, in one frame period, a plurality of images having different exposure times; and
generate image data of the captured plurality of images; and
a recognition unit configured to:
recognize a subject from the generated image data of the plurality of images;
recognize the subject in an image of one frame based on a result of the recognition of the subject from the generated image data; and
adjust an exposure time of the imaging unit based on a level of reliability of the result of the recognition of the subject.
2. The image recognition device according to claim 1 , wherein
the image recognition device comprises a solid-state imaging device, and
the solid-state imaging device includes the imaging unit and the recognition unit.
3. The image recognition device according to claim 1 , further comprising:
a solid-state imaging device that includes the imaging unit; and
a processor that includes the recognition unit.
4. The image recognition device according to claim 1 , wherein the recognition unit is further configured to execute a deep neural network (DNN) process to recognize the subject from the generated image data.
5. The image recognition device according to claim 4 , wherein the recognition unit is further configured to execute the DNN process on the generated image data of the plurality of images in time series.
6. The image recognition device according to claim 4 , wherein the recognition unit is further configured to execute the DNN process on the generated image data of the plurality of images in parallel.
7. The image recognition device according to claim 4 , wherein the recognition unit is further configured to execute the DNN process based on a DNN specialized for each image of the plurality of images having different exposure times.
8. The image recognition device according to claim 1 , wherein the recognition unit is further configured to repeat the adjustment of the exposure time of the imaging unit until the level of reliability of the result of the recognition of the subject converges to a maximum value.
9. The image recognition device according to claim 1 , further comprising a light receiving region that includes a plurality of light receiving elements, wherein
the plurality of light receiving elements have different light receiving areas, and
the imaging unit is further configured to capture the plurality of images by the plurality of light receiving elements.
10. The image recognition device according to claim 1 , further comprising a high dynamic range (HDR) combination unit configured to generate image data of an HDR image by HDR-combining the generated image data of the plurality of images, wherein
the recognition unit is further configured to recognize the subject from the generated image data of the plurality of images before the HDR combination of the generated image data.
11. A solid-state imaging device, comprising:
an imaging unit configured to:
capture, in one frame period, a plurality of images having different exposure times; and
generate image data of the captured plurality of images; and
a recognition unit configured to:
recognize a subject from the generated image data of the plurality of images;
recognize the subject in an image of one frame based on a result of the recognition of the subject from the generated image data; and
adjust an exposure time of the imaging unit based on a level of reliability of the result of the recognition of the subject.
12. An image recognition method, comprising:
capturing, by an imaging unit, a plurality of images having different exposure times in one frame period;
generating image data of the captured plurality of images;
recognizing a subject from the generated image data of the plurality of images;
recognizing the subject in an image of one frame based on a result of the recognition of the subject from the generated image data; and
adjusting an exposure time of the imaging unit based on a level of reliability of the result of the recognition of the subject.