Multi-Scale Image Object Recognition via Feature Vector Fusion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current image processing tools face low recognition accuracy due to coarse boundary estimation, which leads to missing information of the target object in images.
Innovation Solution
An image object recognition method that uses a device with a processor to obtain an instruction feature vector and an image feature vector set, fusing them to recognize the target object in a target image, avoiding coarse boundary estimation by incorporating feature vectors in different scales.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If coarse boundary estimation is provided for the region including the target object, then the processing speed is improved, but the recognition accuracy deteriorates due to information missing
Solution Approach 1:
The patent segments the image feature extraction into multiple scales (first scale, second scale, etc.), where each scale provides feature vectors at different levels of detail. This segmentation allows the system to process images efficiently at coarser scales while capturing fine details at finer scales, thereby resolving the contradiction between processing speed and recognition accuracy.
Solution Approach 2:
The patent introduces a multi-scale dimension by extracting image features at different scales. Instead of relying on a single-scale feature vector, the system incorporates feature vectors from multiple scales, adding a dimensional aspect that enriches the feature representation and improves recognition accuracy without sacrificing processing efficiency.
2Measurement precision
If multiple scales of image feature vectors are used to improve recognition accuracy, then the measurement precision is improved, but the device complexity increases
Solution Approach 1:
The patent employs a universal image processing apparatus that can perform multiple functions: it can extract image features at different scales, generate instruction feature vectors, and integrate these features for recognition. This multi-functional design allows the system to achieve high recognition accuracy without requiring separate dedicated systems for each function, thereby managing device complexity effectively.
Solution Approach 2:
The patent merges the extraction of image feature vectors at multiple scales into a unified processing framework. Instead of using separate systems for each scale, the apparatus combines the multi-scale feature extraction, instruction feature vector generation, and recognition processes into a single integrated system, reducing overall device complexity while maintaining high recognition accuracy.
Data Source
AI summary
The present disclosure describes methods, devices, and storage medium for recognizing a target object in a target image. The method including obtaining, by a device, an image recognition instruction, the image recognition instruction carrying object identification information used for indicating a target object in a target image. The device includes a memory storing instructions and a processor in communication with the memory. The method includes obtaining, by the device, an instruction feature vector matching the image recognition instruction; obtaining, by the device, an image feature vector set matching the target image, the image feature vector set comprising an ith image feature vector for indicating an image feature of the target image in an ith scale, and i being a positive integer; and recognizing, by the device, the target object from the target image according to the instruction feature vector and the image feature vector set.