Computer Vision Object Motion Detection Using Semantic Clustering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing mapping and navigation service providers face challenges in extracting useful and lightweight information from real-time sensor data and inferred scene data for supporting location-based services like intelligent transportation systems.
Innovation Solution
A method and apparatus using computer vision to process video sequences from a vehicle's perspective, detecting semantic localization features, clustering objects, creating object vectors, computing angles with reference vectors, and classifying object motions based on these angles.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If real-time sensor data and inferred scene data are processed using computer vision for advanced driving assistance systems, then detection precision and reliability are improved, but data processing complexity and computational resources increase tremendously
Solution Approach 1:
The patent segments the complex scene data into discrete object detections with semantic labels and localization features. Each object is processed independently through clustering algorithms that group objects by type and spatial relationship, dividing the tremendous data volume into manageable clusters that can be analyzed separately rather than as a monolithic dataset
Solution Approach 2:
The patent extracts only the essential features from raw sensor data - semantic localization features that identify what objects are present and where they are located. This extraction process filters out redundant information and retains only the critical data elements needed for motion detection, significantly reducing processing complexity while maintaining detection precision
2Measurement precision
If high-precision object detection is performed across multiple video frames to support intelligent transportation systems, then object localization accuracy is improved, but processing time and computational load increase
Solution Approach 1:
The patent performs preliminary clustering of objects across multiple frames before conducting detailed motion analysis. By pre-grouping objects into clusters based on their spatial and semantic characteristics in earlier processing stages, the system prepares the data structure in advance, so that subsequent motion detection operations can be performed more efficiently on already-organized clusters rather than raw individual objects
Solution Approach 2:
The patent applies motion detection algorithms selectively to clustered objects rather than processing every detected object individually. By focusing computational resources on representative samples from each cluster and using clustering to infer properties of grouped objects, the system achieves sufficient localization accuracy without the excessive processing time that would result from analyzing every single object in detail
3Reliability
If comprehensive scene data is collected and processed to detect object motions, then detection reliability is improved, but the amount of data to be processed increases tremendously
Solution Approach 1:
The patent merges multiple object detections into unified clusters based on semantic localization features. Objects of the same type in proximity are combined into single clusters, reducing the total number of discrete data elements that need to be processed. This merging maintains detection reliability by preserving information about all objects within each cluster while representing them as a grouped entity
Solution Approach 2:
The patent creates a universal clustering framework that handles multiple object types and scenarios through a single processing mechanism. The same clustering algorithms and motion detection methods apply across diverse object categories and environmental conditions, allowing the system to process comprehensive scene data with a unified approach rather than requiring separate processing pipelines for different data types
Data Source
AI summary
An approach is provided for computer-vision-based object motion detection. The approach involves receiving a video sequence captured from a perspective of a vehicle/device traveling at street level. The approach also involves processing frames of the videos sequence using computer vision to determine semantic localization feature(s) associated with object(s) detected in the frames. The approach further involves clustering the object(s) into object cluster(s) across the frames based on the semantic localization feature(s). The approach further involves creating an object vector connecting the object(s) in a given cluster of the object cluster(s) across the frames. The approach further involves computing an angle between the object vector and a reference vector associated with a movement of the vehicle/device across the frames. The approach further involves classifying a motion of the object(s) relative to the vehicle/device based on the angle.