Region of Interest Video Preprocessing via Runtime Configuration
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current practices in video processing relay the entire frame of view, which is resource and time-consuming due to the need to transfer large, high-resolution video data, as they do not efficiently isolate and preprocess regions of interest from streaming video.
Innovation Solution
A system and method that preprocesses streaming video data using a runtime configuration file to extract and format a region of interest, applying parameters such as cropping, grayscale adjustment, and image processing techniques, and publishes it to an endpoint for further processing by subscribers, allowing real-time adjustments and optimization.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If the entirety of streaming video data (entire frame/field of view) is relayed to the consumer, then complete video information is provided, but processing time and resource consumption increase significantly
Solution Approach 1:
The patent segments the video stream into multiple regions of interest (ROIs) based on detected objects or areas of concern. Instead of processing the entire frame, only the segmented ROI portions are extracted and relayed to consumers, reducing data volume while maintaining relevant information completeness.
Solution Approach 2:
The system extracts only the necessary region of interest from the full video frame using object detection and bounding box identification. This extraction process isolates relevant portions (e.g., detected objects, motion areas) and discards redundant background areas, reducing processing time while preserving critical video information.
2Area of stationary object
If high resolution video data with wider field of view is captured, then more area can be shown, but data transfer requirements and processing resources increase
Solution Approach 1:
The wide field of view is segmented into multiple smaller regions of interest based on detected objects or areas requiring monitoring. Each ROI is processed and transmitted separately, reducing the total data transfer requirements while maintaining comprehensive area coverage through selective focus on relevant zones.
Solution Approach 2:
The system applies different processing qualities to different regions of the video frame. High-resolution processing is applied only to regions containing objects of interest, while other areas receive reduced processing or are excluded entirely, optimizing resource usage while maintaining effective monitoring coverage.
3Reliability
If preprocessing operations are applied to the entire video frame, then comprehensive processing is achieved, but processing time and computational resources increase
Solution Approach 1:
The system extracts only the region of interest from the full video frame before applying preprocessing operations. By performing cropping, resizing, and other preprocessing steps only on the extracted ROI rather than the entire frame, processing efficiency is significantly improved while maintaining processing completeness for the relevant areas.
Solution Approach 2:
The preprocessing pipeline is dynamically configured based on the detected region of interest and the specific requirements of different consumers. The system adapts preprocessing parameters (such as resolution, format, and processing intensity) according to the ROI characteristics and consumer needs, optimizing the balance between processing completeness and efficiency.
Data Source
AI summary
A method of processing first video data from incoming video data includes receiving the incoming video data; preprocessing the incoming video data, by a first computer processor, according to preprocessing parameters, wherein the preprocessing parameters include formatting the incoming video data to create first video data; publishing the first video data to an endpoint; subscribing, by a second computer processor, to the endpoint; and processing, by a second computer processor, the first video data to determine at least one output that is indicative of a first inference dependent upon the first video data. The preprocessing parameters that format the incoming video data to create the first video data are dependent upon the processing to be performed on the first video data.