Face Detection Bounding Box Selection via Object Prioritization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Computer-implemented face detection in video surveillance systems is computationally expensive, especially for real-time identification, as it requires processing entire video frames, which is time-consuming and resource-intensive.
Innovation Solution
A method that selects a cropped bounding box from a video frame based on prioritized characteristics such as trajectory, human classification, skin tone, pixel location, and sharpness, to focus face detection operations on high-priority areas, reducing computational overhead by processing only the most relevant frames.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If face detection operation is performed on the entire video frame, then face detection accuracy is improved, but computational overhead and processing time increase significantly
Solution Approach 1:
The patent divides the video frame into multiple regions of interest (ROIs) based on object detection results, rather than processing the entire frame. This segmentation allows the face detection algorithm to focus only on relevant areas, reducing computational overhead while maintaining detection accuracy.
Solution Approach 2:
The patent applies different processing strategies to different regions of the video frame. High-priority regions (those containing detected objects) receive full face detection processing, while low-priority regions are either processed with simplified algorithms or skipped entirely, optimizing the balance between accuracy and processing time.
2Reliability
If face detection operation is performed on the entire video frame, then comprehensive face detection is achieved, but computational resources are excessively consumed
Solution Approach 1:
The patent performs preliminary object detection and prioritization before executing the face detection operation. By identifying and ranking regions of interest in advance, the system prepares a filtered list of candidate areas, ensuring that comprehensive face detection is performed only where necessary, thus reducing computational resource consumption.
Solution Approach 2:
The patent applies face detection processing selectively to only those regions that meet certain priority thresholds, rather than applying it uniformly across the entire frame. This partial action approach maintains reliability for high-priority regions while significantly reducing overall computational resource usage.
3Productivity
If object prioritization based on multiple characteristics is implemented, then processing efficiency is improved, but system complexity increases
Solution Approach 1:
The patent evaluates multiple characteristics (trajectory, human classification confidence, skin tone presence, pixel sharpness) and transforms them into a unified priority score for each detected object. This parameter transformation approach allows complex multi-factor evaluation to be condensed into a single prioritization metric, improving processing efficiency while managing system complexity through systematic parameter integration.
Data Source
AI summary
A cropped bounding box selection operation is performed on a video captured by a video capture and playback system, to select one or more cropped bounding boxes from the video for processing by a face detection operation. The cropped bounding box selection operation identifies objects from the video images and assigns a ranking to each identified object based on certain priority criteria; one or more cropped bounding boxes corresponding to the objects with the highest ranking(s) are then processed by the face detection operation to detect a face in each object.