Convolutional Neural Network Target Tracking via Euclidean Distance
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for locating tracked targets in video monitoring are inefficient and inaccurate, particularly due to the need for manual input of target position information, which often results in background interference and reduced tracking efficiency.
Innovation Solution
A method utilizing a convolution neural network model with two convolution layers and two fully connected layers to predict target regions in a video frame, calculating Euclidean distances between targets and the original target point, and selecting the target with the smallest distance for tracking, thereby improving accuracy and efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual position information is used for target tracking, then the system is simple to operate, but the locating accuracy is low due to background interference
Solution Approach 1:
The patent replaces the manual mechanical operation of drawing position information with an automated computer vision system using convolutional neural networks. The system automatically identifies target objects in video frames, extracts their position coordinates, and initiates tracking without manual intervention, thereby improving locating accuracy while eliminating background interference issues.
Solution Approach 2:
The system enables self-service by allowing the target object itself to provide its position information through automated detection. The convolutional neural network automatically identifies and extracts position coordinates of targets in video frames, eliminating the need for manual annotation and reducing background interference, thus improving locating accuracy.
2Productivity
If file-based initialization is used, then the system is easy to implement, but the efficiency of acquiring initialized target information is low due to long operation period
Solution Approach 1:
The patent applies preliminary action by pre-training convolutional neural network models with large amounts of target data before actual tracking. This pre-processing enables the system to quickly and accurately identify targets and acquire position information when tracking starts, eliminating the need for lengthy file operations and significantly improving the efficiency of acquiring initialized target information.
Solution Approach 2:
The system replaces the mechanical file reading and parsing process with an automated neural network-based target detection system. The convolutional neural network directly processes video frames to extract target position information, eliminating the time-consuming file I/O operations and significantly reducing the operation period for acquiring initialized target information.
3Measurement precision
If manual drawing of target position is used, then the system is simple to operate, but the locating accuracy is low because large amounts of background information are contained
Solution Approach 1:
The patent replaces manual drawing operations with automated convolutional neural network-based target detection. The system automatically identifies target objects in video frames, extracts precise position coordinates, and eliminates the need for manual drawing, thereby improving position information accuracy by excluding background interference while maintaining ease of operation through automation.
Solution Approach 2:
The system applies the extraction principle by using convolutional neural networks to isolate and extract only the relevant target information from video frames. The network automatically distinguishes target objects from background elements, extracting precise position coordinates without including unnecessary background information, thus improving position information accuracy.
4Productivity
If manual target position input is used, then the system is simple to implement, but the tracking efficiency is reduced due to increased difficulty in tracking
Solution Approach 1:
The patent replaces manual target position input with an automated convolutional neural network system that detects and tracks target objects in real-time. The system automatically processes video frames, identifies targets, calculates position coordinates, and maintains tracking without manual intervention, significantly improving tracking efficiency while managing system complexity through modular architecture.
Solution Approach 2:
The system enables self-service by allowing the tracking system to automatically acquire and process target position information without manual input. The convolutional neural network autonomously identifies targets in video frames, extracts position coordinates, and maintains tracking, thereby improving tracking efficiency while eliminating the need for manual operations.
Data Source
AI summary
A method for tracking a target includes: acquiring original position information of an original target point selected by a user contained in a locating request if the locating request for tracking a target is received; carrying out target prediction on a current frame image according to a preset target prediction model to obtain a target prediction result; calculating an Euclidean distance between each of the targets to be tracked and the original target point according to the target position information and original coordinates of each of the target regions to obtain N distances; selecting a distance with the smallest numerical value from the N distances as a target distance, acquiring target position information corresponding to the target distance, and determining a target to be tracked in a target region corresponding to the obtained target position information as a tracked target corresponding to an original target point.