US 11,915,430 B2
Image analysis apparatus, image analysis method, and storage medium to display information representing flow quantity
Kotaro Yano, Tokyo (JP); Hajime Muta, Zama (JP); and Yasuo Bamba, Tokyo (JP)
Assigned to CANON KABUSHIKI KAISHA, Tokyo (JP)
Filed by CANON KABUSHIKI KAISHA, Tokyo (JP)
Filed on Sep. 9, 2020, as Appl. No. 17/016,067.
Claims priority of application No. 2019-167189 (JP), filed on Sep. 13, 2019.
Prior Publication US 2021/0082127 A1, Mar. 18, 2021
Int. Cl. G06T 7/215 (2017.01); G06T 3/40 (2006.01); G06T 7/269 (2017.01); G06F 3/14 (2006.01)
CPC G06T 7/215 (2017.01) [G06T 3/40 (2013.01); G06T 7/269 (2017.01); G06F 3/14 (2013.01); G06T 2207/20021 (2013.01); G06T 2207/20084 (2013.01); G06T 2207/20092 (2013.01); G06T 2207/30196 (2013.01); G06T 2207/30242 (2013.01)] 20 Claims
OG exemplary drawing
 
1. An image processing apparatus comprising:
one or more processors; and
one or more memories storing instructions that, when executed by the one or more processors, cause the one or more processors to be configured to operate as:
an acquisition unit configured to acquire time-series images captured by an image sensor;
a setting unit configured to set a direction of a detection line for measuring a flow quantity of targets, and set a direction of measuring the flow quantity of the targets in the time-series images in a direction intersecting the direction of the detection line;
a flow quantity estimation unit configured to estimate information representing the flow quantity of the targets based on a plurality of images acquired from the time-series images and the direction of measuring the flow quantity;
a passage count estimation unit configured to measure a number of people passing the detection line set by the setting unit based on the flow quantity of the targets estimated by the flow quantity estimation unit; and
a display control unit configured to control a display to display the information representing the flow quantity of the targets and the number of people passing the detection line that is estimated based on the plurality of images acquired from the time-series images and the direction of measuring the flow quantity of the targets, together with information indicating the direction of measuring the flow quantity of the targets,
wherein the flow quantity estimation unit estimates the information representing the flow quantity of the targets with a convolutional neural network having convolutional layers and deconvolutions layers, and
wherein the neural network is configured to receives the information on the direction of measuring the flow quantity of the targets, and determine a weight coefficient for the convolution layers based on the information on the direction of measuring the flow quantity of the targets, at least one of the convolution layers generating a feature map, and
wherein a deconvolution layer is selected among the deconvolution layers to perform a deconvolution on the feature map to generate the flow quantity.