Hybrid GPU-FPGA System for Secure Person Tracking

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Conventional techniques for person tracking and data privacy in autonomous machines are limited by occlusions, low lighting conditions, and the insecurity of edge devices, particularly in scenarios like indoor environments, and fail to provide superior outputs for facial recognition and acceleration of vision-based algorithms.

Innovation Solution

A novel technique using adaptive color histogram-based object tracking, motion detection, and pose estimation for human body tracking, combined with secure enclaves for deep learning workloads to ensure data privacy, and a hybrid GPU-FPGA system for accelerating data processing in autonomous driving.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If conventional object tracking techniques are used, then basic tracking functionality is provided, but tracking performance deteriorates under occlusions and changing lighting conditions

Engineering Contradiction:

Improvetracking performanceVSAvoidocclusions and lighting changes

Core Design Contradiction:

ReliabilityVSObject-affected harmful factors

Solution Approach 1:

The patent segments the object tracking task into multiple components: color histogram analysis, motion detection, and pose estimation. Each component processes specific aspects of the target object independently, allowing the system to maintain tracking reliability even when partial occlusions occur or lighting conditions change, as different segments can compensate for each other's limitations

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent dynamically adjusts tracking parameters based on environmental conditions. The color histogram parameters are adapted to changing lighting conditions, motion detection thresholds are modified based on scene activity levels, and pose estimation parameters are refined according to occlusion detection, enabling the system to maintain high tracking reliability across varying conditions

Inventive Principle:

Principle #35Parameter changes

2Measurement precision

If facial recognition algorithms are enhanced for better accuracy, then recognition precision improves, but processing time and computational resources increase

Engineering Contradiction:

Improvefacial recognition accuracyVSAvoidprocessing time

Core Design Contradiction:

Measurement precisionVSLoss of time

Solution Approach 1:

The patent performs preliminary pose estimation and motion detection before executing full facial recognition algorithms. By pre-processing the data to identify potential targets and their orientations, the system reduces the search space for facial recognition, thereby maintaining high accuracy while significantly reducing processing time

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The patent applies different processing qualities to different regions and stages of facial recognition. High-precision algorithms are applied only to identified candidate regions, while lower-precision but faster methods are used for initial screening and background regions, optimizing the balance between accuracy and processing speed

Inventive Principle:

Principle #3Local quality

3Reliability

If edge devices process raw image, speech, and text data locally, then data privacy is improved, but security vulnerabilities increase

Engineering Contradiction:

Improvedata privacyVSAvoidsecurity vulnerabilities

Core Design Contradiction:

ReliabilityVSObject-affected harmful factors

Solution Approach 1:

The patent introduces secure enclaves as intermediary protected environments within edge devices. These enclaves act as secure containers that isolate sensitive data processing from the rest of the system, allowing local processing benefits while mitigating security risks through hardware-based protection boundaries

Inventive Principle:

Principle #24Intermediary (Mediator)

Solution Approach 2:

The patent implements security measures in advance by establishing secure enclaves before sensitive data processing occurs. These pre-configured protected environments prevent potential security vulnerabilities from being exploited, as the defensive structure is already in place before any data processing or potential attacks occur

Inventive Principle:

Principle #9Preliminary anti-action

4Productivity

If vision-based algorithms are accelerated for autonomous driving, then processing speed improves, but system complexity increases

Engineering Contradiction:

Improvedata processing speedVSAvoidsystem complexity

Core Design Contradiction:

ProductivityVSDevice complexity

Solution Approach 1:

The patent merges multiple specialized processing functions (color histogram computation, motion detection, pose estimation, and secure processing) into a unified hybrid GPU-FPGA system. This consolidation achieves high processing speed through parallel architecture while managing complexity through integrated design and shared resources

Inventive Principle:

Principle #5Merging (Combining)

Data Source

PatentUS11393211B2Hybrid graphics processor-field programmable gate array system

Publication Date: 2022.07.19 INTEL CORP

AI summary

A mechanism is described for facilitating person tracking and data security in machine learning at autonomous machines. A method of embodiments, as described herein, includes detecting, by a camera associated with one or more trackers, a person within a physical vicinity, where detecting includes capturing one or more images the person. The method may further include tracking, by the one or more trackers, the person based on the one or more images of the person, where tracking includes collect tracking data relating to the person. The method may further include selecting a tracker of the one or more trackers as a preferred tracker based on the tracking data.