Image Recognition Using Pixel Estimation and Segmentation Models
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current computer vision technologies face challenges in ensuring the accuracy of target object recognition in images, particularly when there are insufficient historical images for training accurate segmentation models, leading to inefficiencies and reduced precision in recognition processes.
Innovation Solution
A method that combines pixel-level estimation and segmentation model-based recognition, where the location of a target object is estimated using pixel values and location relationships under specific conditions, and inputted to a segmentation model for precise location determination, allowing for accurate recognition regardless of historical image quantity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If only segmentation models are used for target object recognition, then recognition precision can be improved when sufficient historical images are available, but the system becomes unreliable when historical images are insufficient
Solution Approach 1:
The patent performs preliminary pixel-level analysis on the current image to estimate target object location before determining whether to use segmentation model recognition. This preliminary action ensures that the system can always provide a baseline recognition result, improving reliability while allowing segmentation models to enhance precision when conditions permit
Solution Approach 2:
The patent dynamically adjusts the recognition strategy based on the quantity of historical images available. When historical images are sufficient, the system switches to segmentation model-based recognition for higher precision; when historical images are insufficient, it relies on pixel-level estimation, ensuring continuous reliable operation under varying conditions
2Reliability
If pixel-level estimation is used for target object location, then the system can operate with insufficient historical images, but recognition precision is reduced
Solution Approach 1:
Pixel-level estimation is performed as a preliminary step to establish a baseline recognition result when historical images are insufficient. This ensures system reliability and provides a fallback mechanism that maintains operational capability while accepting reduced precision in data-scarce scenarios
Solution Approach 2:
The system dynamically selects between pixel-level estimation and segmentation model recognition based on data availability. Pixel-level estimation is activated when historical images are insufficient, ensuring reliable operation, while the system transitions to more precise segmentation-based methods when conditions allow
3Measurement precision
If a dual recognition approach combining pixel-level estimation and segmentation models is used, then recognition accuracy is improved across varying conditions, but device complexity increases
Solution Approach 1:
The patent implements a dynamic recognition system that adapts its complexity based on conditions. The system evaluates the quantity of historical images and automatically selects the appropriate recognition approach, using simpler pixel-level estimation when needed and more complex segmentation models when beneficial, thereby optimizing the balance between accuracy and complexity
Solution Approach 2:
The system applies different recognition methods to different situations rather than using a uniform approach. Pixel-level estimation is applied when historical data is insufficient, while segmentation models are applied when data is sufficient, ensuring that each local condition is handled by the most appropriate method
Data Source
AI summary
An image recognition method includes: obtaining an image; extracting a target image region corresponding to a target part from the image, wherein the target image region includes a target object; determining a location of the target object in the target image region (i) according to pixel values of pixels in the target image region and a location relationship between the pixels, or (ii) inputting the target image region to a trained segmentation model to obtain the location of the target object in the target image region; and displaying a recognition result of the image, wherein the recognition result indicates the location of the target object in the target image region.