Multi-Target Detection Model for 2D Code Tracking on Small Objects
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing detection methods struggle with efficiently and accurately recognizing miniature and moving 2D codes, especially when attached to small objects like insects, due to their small size and varying orientations, which results in slow detection speeds and limited real-time tracking capabilities.
Innovation Solution
A detection method utilizing a pre-trained multi-target detection model to process images containing 2D codes, allowing for the recognition of 2D codes and the determination of their actual position relative to an imaging device, even under complex conditions such as varying illumination and deformation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If traditional code scanning methods are used to detect 2D codes on small objects, then the detection process can be simple, but the detection speed is slow and real-time tracking capability is limited
Solution Approach 1:
The patent applies preliminary action by pre-training a multi-target detection model with extensive training data before actual detection. This pre-training phase prepares the model to quickly and accurately identify 2D codes on small moving objects during real-time operation, resolving the contradiction between detection speed and accuracy by doing the complex learning work beforehand.
Solution Approach 2:
The patent replaces traditional mechanical code scanning methods with an AI-based multi-target detection model. This substitution enables the system to achieve both high detection speed and high accuracy simultaneously, as the neural network can process images rapidly while maintaining the ability to precisely identify 2D codes even on small, moving objects.
2Measurement precision
If a multi-target detection model is used to detect 2D codes on moving objects, then detection accuracy improves, but the complexity of the detection system increases
Solution Approach 1:
The patent applies universality by designing a multi-target detection model that can simultaneously detect multiple types of objects and 2D codes in a single processing step. This multi-functional approach improves detection accuracy while managing system complexity by consolidating multiple detection tasks into one unified model rather than requiring separate systems for each target type.
3Adaptability or versatility
If the 2D code is made smaller to fit on small objects like insects, then the object can be marked, but the recognition becomes more difficult and slower
Solution Approach 1:
The patent replaces traditional mechanical recognition methods with an AI-based multi-target detection model that is specifically trained to identify small 2D codes on small objects. This substitution enables the system to maintain high recognition speed even when the 2D code size is reduced to fit on small objects like insects, resolving the contradiction between adaptability and productivity.
Solution Approach 2:
The patent applies preliminary action by pre-training the detection model with extensive data including small 2D codes on small objects. This pre-training prepares the model to efficiently recognize miniature codes during real-time operation, maintaining high recognition speed despite the reduced code size and the challenges it presents.
Data Source
AI summary
The present disclosure relates to a detection method and apparatus, an object monitoring system, a computing device, and a storage medium. The detection method includes acquiring an image containing a 2D code for an object, and inputting the image into a pre-trained multi-target detection model configured to detect multiple classes of detection targets including the 2D code. The method further includes at least one of: based on a detection box of the 2D code in the image, cropping out an image region of the determined detection box of the 2D code from the image, and recognizing the 2D code from the cropped image region; or, based on an actual size of the 2D code, a size of the 2D code in the image, and a position of the 2D code in the image, determining an actual position of the 2D code relative to an imaging device.