Multi-object detection ranking and overlap filtering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current non-maximum suppression methods for multi-object detection in computer vision are inefficient for real-time detection of densely distributed targets, particularly for multiple categories, as they often result in missed detections and are computationally complex.
Innovation Solution
An image processing method that ranks detected targets based on parameters, checks category conditions, and determines overlap to selectively remove targets, using thresholds and ratios to refine the detection dataset, improving detection accuracy for multiple categories.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If non-maximum suppression is used to extract valid feature points, then detection accuracy is improved, but computational complexity increases and real-time detection becomes difficult
Solution Approach 1:
The patent segments the non-maximum suppression process into distinct functional modules: confidence score calculation unit, overlap detection unit, and target suppression unit. This segmentation allows each module to perform a specific function efficiently, reducing overall computational complexity while maintaining detection accuracy through systematic processing of detected targets
Solution Approach 2:
The patent performs preliminary ranking of detected targets based on confidence scores before executing the full non-maximum suppression algorithm. By pre-sorting targets in descending order of confidence, the system identifies high-priority targets first, enabling early termination of suppression operations for low-confidence targets and reducing computational burden
2Productivity
If heuristic window fusion algorithm is used for densely distributed multi-object detection, then detection speed is improved, but a large number of missed detections occur
Solution Approach 1:
The patent implements a feedback mechanism where the overlap detection unit continuously monitors the spatial relationship between detected targets and previously identified high-confidence targets. When overlap exceeds a threshold, the system feedbacks to suppress the current target, preventing missed detections while maintaining detection speed through efficient feedback-based decision making
3Measurement precision
If non-maximum of greedy strategies is used for single category detection, then detection accuracy is improved, but it cannot be effectively applied to multiple categories of densely distributed targets
Solution Approach 1:
The patent creates a universal non-maximum suppression framework that handles multiple categories through category-agnostic overlap detection and confidence-based suppression. The system processes targets from different categories using the same suppression logic, making it adaptable to multi-category detection while maintaining the accuracy benefits of greedy strategies through systematic confidence score evaluation
Data Source
AI summary
The present disclosure provides an image processing method, device and apparatus, and a computer-readable storage medium. The image processing method includes: obtaining an image and detecting multiple targets in the image; obtaining categories of multiple detected targets and first parameters of a detection frame corresponding to each detected target, and ranking all detected targets according to the first parameters in a descending order to obtain a dataset; obtaining a target with a highest first parameter in the dataset as a first detected target, and detecting whether the first parameter of a first detection frame corresponding to the first detected target satisfies a preset category condition of a category to which the first detected target belongs; detecting whether the first detection frame corresponding to the first detected target overlaps with a detection frame corresponding to a respective one of the plurality of detected targets except the first detected target.