US 10,380,853 B1Granted

Presence detection and detection localization

AssigneeAmazon Technologies, Inc.

InventorsMashhour Solh, Lelin Zhang, Jinglun Gao

Life of this patent

Estimating the term…

Loading patent…

Abstract

A system configured to improve human presence detection and/or localization by generating aggregate confidence values. The system may aggregate confidence values corresponding to overlapping regions of interest. The system may perform human presence detection by comparing the aggregate confidence values to a universal threshold, with aggregate confidence values above the universal threshold indicating that human presence is detected. The system may use the aggregate confidence values to generate a heatmap, may identify a strongest cluster of pixels in the heatmap and determine a bounding box surrounding the strongest cluster. To distinguish the strongest cluster from a second strongest cluster of pixels, the system may apply a Gaussian kernel with varying threshold values. The system may store the threshold value at which the strongest cluster separates from the second strongest cluster and may use the threshold value to generate the bounding box.

Claims20

Dependent claims

A computer-implemented method for generating a bounding box indicating pixels associated with human presence, comprising:

receiving first image data;
determining a first candidate area in the first image data, the first candidate area including a first plurality of pixels including a first pixel;
determining a first confidence value that a human is represented in the first candidate area;
determining a second candidate area in the first image data, the second candidate area including a second plurality of pixels including the first pixel;
determining a second confidence value that a human is represented in the second plurality of pixels;
determining a first aggregate confidence value corresponding to the first pixel, the first aggregate confidence value including a sum of the first confidence value and the second confidence value;
determining a second aggregate confidence value corresponding to a second pixel in the first image data;
generating pixel intensity data based on aggregate confidence values, the aggregate confidence values including the first aggregate confidence value and the second aggregate confidence value, the pixel intensity data including a first pixel intensity value associated with the first pixel;
determining a peak pixel intensity value having a highest pixel intensity value in the pixel intensity data;
determining, based on the pixel intensity data, a first cluster of pixels including a peak pixel having the peak pixel intensity value;
determining, using a region growth algorithm, a bounding box corresponding to the first cluster of pixels, the bounding box indicating pixels associated with human presence;
and
sending output data indicating that human presence is detected, the output data including the bounding box.

10 more elements

A computer-implemented method, comprising:

receiving first image data;
determining a first portion of the first image data, the first portion including a first plurality of pixels including a first pixel;
determining a first confidence value for the first portion, the first confidence value indicative of a probability of a human being represented in the first portion;
determining a second portion of the first image data, the second portion including a second plurality of pixels including the first pixel;
determining a second confidence value for the second portion, the second confidence value indicating of a probability of a human being represented in the second plurality of pixels;
determining a first aggregate confidence value corresponding to the first pixel, the first aggregate confidence value being a sum of the first confidence value and the second confidence value;
determining a second aggregate confidence value corresponding to a second pixel in the first image data;
determining a peak aggregate confidence value, the peak aggregate confidence value corresponding to a highest aggregate confidence value associated with the first image data;
determining that the peak aggregate confidence value is above a detection threshold value;
determining that human presence is detected in the image data;
and
performing, in response to determining that human presence is detected, an action.

9 more elements

A device comprising:

at least one processor;
and
memory including instructions operable to be executed by the at least one processor to perform a set of actions to configure the device to:
receive first image data;
determine a first portion of the first image data, the first portion including a first plurality of pixels including a first pixel;
determine a first confidence value for the first portion, the first confidence value indicative of a probability of a human being represented in the first portion;
determine a second portion of the first image data, the second portion including a second plurality of pixels including the first pixel;
determine a second confidence value for the second portion, the second confidence value indicating of a probability of a human being represented in the second plurality of pixels;
determine a first aggregate confidence value corresponding to the first pixel, the first aggregate confidence value being a sum of the first confidence value and the second confidence value;
determine a second aggregate confidence value corresponding to a second pixel in the first image data;
determine a peak aggregate confidence value, the peak aggregate confidence value corresponding to a highest aggregate confidence value associated with the first image data;
determine that the peak aggregate confidence value is above a detection threshold value;
determine that human presence is detected in the image data;
and
send output data indicating that human presence is detected.

12 more elements

Dependencies are inferred from the claim text.

Facts

Application: 15/601,752Filed 22 May 2017
Granted: 13 Aug 2019
Term ends: Not estimatedThe source records don’t support an estimate.

Classification14 codes

G06V 40/10
Recognition of biometric, human-related or animal-related patterns in image or video data: human or animal bodies, e.g. vehicle occupants or…
G06F 18/2321
Non-hierarchical techniques: using statistics or function optimisation, e.g. modelling of probability density functions
G06V 10/25
Image preprocessing: determination of region of interest [ROI] or a volume of interest [VOI]
G06V 10/763
Using clustering, e.g. of similar faces in social networks: non-hierarchical techniques, e.g. based on statistics of modelling distributions