Accumulated Classification Scores for Mobile Vision Recognition
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Machine vision systems face challenges in maintaining high classification accuracy, especially in conditions with poor lighting, blurring, or other environmental factors, such as those encountered in mobile device-based object recognition applications.
Innovation Solution
A system that captures multiple images of an object from different views, perspectives, and under varying lighting conditions, using a trained neural network to generate and accumulate classification scores, and outputs enhanced classification information when the accumulated score exceeds a predetermined threshold.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multiple images are captured and classification scores are accumulated, then object recognition accuracy is improved, but processing time and system complexity increase
Solution Approach 1:
The system performs preliminary actions by capturing multiple images and accumulating classification scores before making a final recognition decision. This allows the system to build confidence in the classification result progressively, improving accuracy while providing a mechanism to stop when the threshold is reached, thus managing processing time.
Solution Approach 2:
The system uses feedback from accumulated classification scores to determine when to stop processing and output a result. By continuously monitoring the accumulated score against a predetermined threshold, the system can dynamically adjust processing duration, stopping early when sufficient confidence is achieved, thereby balancing accuracy improvement with time efficiency.
2Measurement precision
If multiple images are captured and classification scores are accumulated, then object recognition accuracy is improved, but device complexity increases
Solution Approach 1:
The system merges multiple classification scores from different images into a single accumulated score. This combining approach allows the system to leverage information from multiple images without requiring entirely separate processing pipelines, thus improving accuracy while controlling the increase in system complexity through a unified accumulation mechanism.
Solution Approach 2:
The classification engine is designed to be universal and multi-functional, capable of processing multiple images and accumulating scores using the same core algorithms and data structures. This multi-functionality allows the system to handle varied input conditions (different lighting, angles, qualities) without requiring specialized processing paths for each scenario, thereby improving robustness while limiting complexity growth.
3Productivity
If single image processing is used, then processing speed is maintained, but recognition accuracy degrades in poor lighting and environmental conditions
Solution Approach 1:
The system applies partial action by processing only as many images as needed to reach the classification threshold. In good conditions, fewer images are processed (partial action), maintaining speed. In poor conditions, more images are processed (excessive action relative to single-image processing), improving accuracy. This adaptive approach balances processing speed with classification accuracy based on environmental conditions.
Data Source
AI summary
The disclosed technology includes systems and methods for enhancing machine vision object recognition based on a plurality of captured images and an accumulation of corresponding classification analysis scores. A method is provided for capturing, with a camera of a mobile computing device, a plurality of images, each image of the plurality of images comprising a first object. The method includes processing, with a classification module comprising a trained neural network processing engine, at least a portion of the plurality of images. The method includes generating, with the classification module and based on the processing, one or more object classification scores associated with the first object. The method includes accumulating, with an accumulating module, the one or more object classification scores. And responsive to a timeout or an accumulated score exceeding a predetermined threshold, the method includes outputting classification information of the first object.