Method and apparatus for checkout based on image identification technique of convolutional neural network

A checkout apparatus based on an image identification technique of convolutional neural network, comprising:

• an acquisition device configured to acquire N pictures of M items to be classified placed on a checkout counter, the N pictures corresponding to N shooting angles one to one, the N shooting angles have at least one first shooting angle, and the first shooting angle is the shooting angle for shooting the M items downwards from a position just above the M items;
• an object detection device configured to firstly perform object detection in the picture acquired at the first shooting angle to obtain M first rectangular area images corresponding to the M items one to one, and perform the object detection in the remaining pictures of the N pictures respectively, to obtain M remaining rectangular area images corresponding to the M items one to one in each of the pictures, according to the number of the first rectangular area images;
• a first classification device configured to obtain N preliminary classification results corresponding to each of the items according to the first rectangular area images, the remaining rectangular area images and a pre-trained first-level classification model, and obtain a first-level classification result corresponding to each of the items according to the N preliminary classification results corresponding to each of the items and a pre-trained first-level linear regression model, the first-level classification model being a model that is constructed by the image identification technique of convolutional neural network and trained by all the items in the shopping place;
• a confirmation device configured to use the first-level classification result corresponding to each of the items as a first classification result of each of the items;
• and
• a checkout device configured to perform checkout of the M items according to the first classification result of each of the items,
• M and N being natural numbers greater than or equal to 2.

A checkout apparatus based on an image identification technique of convolutional neural network, comprising:

• N cameras configured to acquire N pictures of M items to be classified placed on a checkout counter, the N pictures corresponding to N shooting angles one to one, the N shooting angles have at least one first shooting angle, and the first shooting angle is the shooting angle for shooting the M items downwards from a position just above the M items;
• a processor;
• and
• a memory configured to record processor-executable instructions,
• the processor configured to
• perform object detection in the picture acquired at the first shooting angle firstly to obtain M first rectangular area images corresponding to the M items one to one,
• perform the object detection in the remaining pictures of the N pictures respectively, to obtain M remaining rectangular area images corresponding to the M items one to one in each of the pictures, according to the number of the first rectangular area images,
• obtain N preliminary classification results corresponding to each of the items according to the first rectangular area images, the remaining rectangular area images and a pre-trained first-level classification model,
• obtain a first-level classification result corresponding to each of the items according to the N preliminary classification results corresponding to each of the items and a pre-trained first-level linear regression model, the first-level classification model being a model that is constructed by the image identification technique of convolutional neural network and trained by all the items in the shopping place,
• use the first-level classification result corresponding to each of the items as a first classification result of each of the items, and
• perform checkout of the M items according to the first classification result of each of the items,
• M and N being natural numbers greater than or equal to 2.