Object Type Prediction via Dual-Image Feature Comparison

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Current image recognition methods face challenges in accurately predicting object types due to background interference and information loss, particularly when the target object occupies a small area in the image, leading to reduced accuracy and false detection rates.

Innovation Solution

Combining full image recognition techniques with subject area detection methods to process both full image feature data and subject area feature data, comparing results to determine the accurate type of the object, thereby reducing background interference and information loss.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Area of stationary object

If full image feature data is used for object recognition, then the recognition process can be performed on the entire image, but background information is introduced that interferes with target object recognition, leading to lower accuracy

Engineering Contradiction:

Improveimage processing areaVSAvoidobject type prediction accuracy

Core Design Contradiction:

Area of stationary objectVSMeasurement precision

Solution Approach 1:

The patent divides the image processing into two segments: full image processing to obtain first type prediction result, and subject area processing to obtain second type prediction result. By segmenting the processing areas and combining results, the method reduces background interference while maintaining comprehensive object recognition capability.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent extracts the subject area from the full image through subject area detection. By taking out only the relevant subject region for separate processing and comparison with full image results, the method eliminates interfering background information while preserving target object recognition accuracy.

Inventive Principle:

Principle #2Taking out (Extraction)

2Object-affected harmful factors

If only subject area is analyzed for object recognition, then background interference is reduced, but scene information and context information are lost, reducing classification accuracy when object characteristics are similar

Engineering Contradiction:

Improvebackground interferenceVSAvoidobject type classification accuracy

Core Design Contradiction:

Object-affected harmful factorsVSMeasurement precision

Solution Approach 1:

The patent merges the advantages of both full image processing and subject area processing by obtaining type prediction results from both approaches and comparing them. The final object type is determined when both results match, combining the background reduction benefit of subject area analysis with the context preservation benefit of full image analysis.

Inventive Principle:

Principle #5Merging (Combining)

Solution Approach 2:

The patent uses feedback by comparing the first type prediction result (from full image) with the second type prediction result (from subject area). This comparison acts as a verification mechanism that feedbacks into the final determination, ensuring both background reduction and context utilization for accurate classification.

Inventive Principle:

Principle #23Feedback

3Productivity

If subject area detection is used to identify objects, then the processing is more focused on the target, but false detection rate exists in the subject area detection algorithm, introducing loss into prediction results

Engineering Contradiction:

Improveobject recognition efficiencyVSAvoidprediction result accuracy

Core Design Contradiction:

ProductivityVSReliability

Solution Approach 1:

The patent performs preliminary subject area detection to identify the region of interest before final classification. This preliminary action enables focused processing on the target area, improving efficiency while the subsequent comparison with full image results corrects potential false detections, maintaining reliability.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The patent cushions against false detection errors by having a verification mechanism in place beforehand. The comparison between full image type prediction and subject area type prediction serves as a protective measure that prevents false detections from affecting the final result, ensuring reliability.

Inventive Principle:

Principle #11Beforehand cushioning (Prior cushioning)

Data Source

PatentUS10706334B2Type prediction method, apparatus and electronic device for recognizing an object in an image

Publication Date: 2020.07.07 ALIBABA GROUP HOLDING LTD

AI summary

Type prediction method, apparatus and electronic device for recognizing an object in an image are disclosed. The method may include processing an image to be processed using a full image recognition technique to obtain a first type prediction result of an object in the image to be processed; processing a subject area of the image to be processed using a feature recognition technique to obtain a second type prediction result of an object of the subject area; determining whether the first type prediction result matches the second type prediction result; if the first type prediction result matches the second type prediction result, determining a type of the object of the image to be processed to be the first type prediction result or the second type prediction result.