Mobile apparatus with computer vision elements for product identifier detection with minimal detection adjustments
Disclosed are systems and techniques for identifying product identifiers in images. A technique can include receiving, by an edge computing device, x and y coordinates for a location of an out of stock shelf section, determining a frame of reference (FOR) based on the coordinates, the FOR including a location of a product identifier for the out of stock shelf section, determining incremental adjustments to a camera based on the FOR, instructing the camera to adjust by the incremental adjustments and capture an image of the location of the product identifier, performing image analysis on the image to identify the product identifier, determining second incremental adjustments to the camera to focus the camera on the location of the product identifier if the product identifier is not identifiable in the image, and instructing the camera to reset to original camera settings if the product identifier is identifiable in the image.
1 . A system for identifying items in a retail environment, the system comprising:
a housing configured to attach to a cart, the cart being moved throughout the retail environment, wherein the housing comprises:
a first camera configured to capture first images of a space surrounding the cart as the cart is moved throughout the retail environment, and
a second camera that is different from the first camera and configured to capture second images of a particular area in the space surrounding the cart, wherein the particular area in the space surrounding the cart comprises a subset of a field of view of the first camera; and
a controller comprising one or more processors and memory storing instructions that, when executed, cause the one or more processors to perform operations comprising:
receiving the first images from the first camera;
identifying an out of stock condition in the space surrounding the cart based on processing the first images;
instructing the second camera to iteratively capture the second images of the particular area in the space surrounding the cart in response to identifying the out of stock condition;
receiving the second images from the second camera; and
identifying a product in the particular area having the out of stock condition based on processing the second images;
wherein the controller comprises an edge computing device configured to attach to the housing.
2 . The system of claim 1 , wherein processing the second images comprises identifying a product identifier of the product in the second images using object detection techniques.
3 . The system of claim 1 , wherein the out of stock condition is identified based on identifying a threshold area of the first images that resembles a shelf feature in the space surrounding the cart.
4 . The system of claim 1 , wherein the controller is further configured to identify the particular area in the space surrounding the cart that corresponds to a location of the identified out of stock condition in the first images.
5 . The system of claim 1 , wherein the controller is further configured to:
predict a location of the second camera relative to the particular area in the space surrounding the cart;
determine a field of view of the second camera at the predicted location;
determine incremental adjustments to the second camera that align the field of view of the second camera with a frame of reference that corresponds to the particular area in the space surrounding the cart; and
transmit the incremental adjustments to the second camera for execution as the cart moves throughout the retail environment relative to the predicted location, wherein the second camera is configured to capture the second images in response to execution of the incremental adjustments.
6 . An apparatus for identifying items in a retail environment, the apparatus comprising:
a housing configured to attach to a portion of a cart, the cart being moved throughout the retail environment, wherein the housing comprises a vertical pole;
a first camera attached to the housing and configured to capture first images of a space surrounding the cart as the cart is moved throughout the retail environment; and
a second camera that is different from the first camera and attached to the housing, wherein the second camera is configured to capture second images of a particular area in the space surrounding the cart based on an identification of a shelf feature in the first images, wherein the particular area in the space surrounding the cart comprises the shelf feature, wherein the particular area in the space surrounding the cart is a subset of a viewpoint of the first camera.
7 . The apparatus of claim 6 , wherein the apparatus further comprises a controller configured to:
receive the first images;
process the first images to determine whether a threshold area in the first images resembles the shelf feature;
instruct the second camera to iteratively capture the second images in response to determining that the threshold area in the first images resembles shelf feature;
receive the second images; and
process the second images.
8 . The apparatus of claim 7 , wherein processing the second images comprises determining an out of stock condition of products that are detected in the second images.
9 . The apparatus of claim 7 , wherein processing the second images comprises identifying each product in the second images based on detecting a corresponding product identifier using object detection techniques.
10 . The apparatus of claim 6 , wherein the first camera is a fixed camera.
11 . The apparatus of claim 6 , wherein the second camera is a moveable camera.
12 . The apparatus of claim 6 , wherein the first camera is a low resolution camera and the second camera is a high resolution camera.
13 . The apparatus of claim 6 , wherein the housing is configured to attach to a front corner of the cart.
14 . The apparatus of claim 6 , wherein the apparatus further comprises a light attached to the housing.
15 . The apparatus of claim 14 , wherein the light is configured to be selectively actuated to illuminate an area that is being imaged by the first camera or the second camera.
16 . The apparatus of claim 6 , wherein the apparatus further comprises an output indicator attached to the housing and configured to be selectively illuminated in response to determinations being made by a controller of the apparatus.
17 . An apparatus for identifying items in a retail environment, the apparatus comprising:
a housing configured to attach to a portion of a cart, the cart being moved throughout the retail environment;
a first camera attached to the housing and configured to capture first images of a space surrounding the cart as the cart is moved throughout the retail environment;
a second camera that is different from the first camera and attached to the housing, wherein the second camera is configured to capture second images of a particular area in the space surrounding the cart based on an identification of a shelf feature in the first images, wherein the particular area in the space surrounding the cart comprises the shelf feature, wherein the particular area in the space surrounding the cart is a subset of a viewpoint of the first camera; and
an indoor location device attached to the housing, positioned at a top of the housing, and facing a ceiling of the retail environment, wherein the indoor location device is configured to detect a location of the cart based on identifying lights in the ceiling of the retail environment as the cart is being moved throughout the retail environment.
18 . The apparatus of claim 17 , wherein the apparatus further comprises a controller configured to:
receive the first images;
process the first images to determine whether a threshold area in the first images resembles the shelf feature;
instruct the second camera to iteratively capture the second images in response to determining that the threshold area in the first images resembles shelf feature;
receive the second images; and
process the second images.
19 . The apparatus of claim 18 , wherein processing the second images comprises determining an out of stock condition of products that are detected in the second images.
20 . The apparatus of claim 18 , wherein processing the second images comprises identifying each product in the second images based on detecting a corresponding product identifier using object detection techniques.