Attention Module Item Identification Accuracy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing machine learning models for identifying items in images struggle with accuracy due to interference from other items in the image, especially when the camera angle changes, as they treat all areas equally without distinguishing importance.
Innovation Solution
A method and apparatus that utilize a pre-trained attention module and long short-term memory network to focus on the to-be-identified item by setting initial position coordinates, iteratively refining the identification through a pre-trained attention module and long short-term memory network, and segmenting the item image using pre-trained target detection models.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If a machine learning model treats all areas in the item image with equal importance, then the model structure is simple, but the identification accuracy deteriorates due to interference from other items
Solution Approach 1:
The patent applies local quality by making different parts of the image have different weights through the attention mechanism. The attention module calculates attention weights for different regions, allowing the model to focus computational resources on relevant areas while suppressing irrelevant ones, thereby resolving the contradiction between simple structure and high accuracy
2Measurement precision
If a pre-trained attention module and long short-term memory network are used to focus on the to-be-identified item, then the identification accuracy is improved, but the device complexity increases
Solution Approach 1:
The patent uses pre-trained attention modules and long short-term memory networks that have been trained in advance on large datasets. This preliminary action allows the model to acquire general knowledge about item recognition, reducing the need for complex architecture design while maintaining high identification accuracy
Data Source
AI summary
Embodiments of the present disclosure provide a method and apparatus for identifying an item. The method includes: acquiring an item image of a to-be-identified item; setting initial position coordinates of the to-be-identified item on the item image; and executing following identifying: inputting the item image and the initial position coordinates into a pre-trained attention module to output an item feature of the to-be-identified item; inputting the item feature into a pre-trained long short-term memory network to output a predicted category and predicted position coordinates of the to-be-identified item; determining whether a preset condition is satisfied; and determining, in response to the preset condition being satisfied, a predicted category of the to-be-identified item outputted by the long short-term memory network a last time for use as a final category of the to-be-identified item.