Vehicle Recognition Using Deep Learning Brand and Color Extraction

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current vehicle recognition technologies face challenges in accurately identifying vehicles when license plates are obscured, stained, or at a distance, leading to low efficiency and high error rates due to reliance on human judgment and traditional image processing methods.

Innovation Solution

A vehicle recognition system utilizing deep learning models, such as convolutional neural networks (CNN), deep belief networks (DBN), and recurrent neural networks (RNN), to extract and recognize vehicle brand and color features from images, prioritizing the vehicle face as the primary feature for brand recognition and employing preprocessing techniques to enhance image quality.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If traditional image processing methods are used for vehicle recognition, then the system is simple to implement, but recognition accuracy decreases when license plates are obscured, stained, or at a distance

Engineering Contradiction:
Improvevehicle recognition accuracyVSAvoidsystem complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent replaces traditional mechanical/image processing methods with deep learning neural networks. Specifically, it uses convolutional neural networks (CNN) for feature extraction and recurrent neural networks (RNN) for sequence recognition, substituting conventional computer vision algorithms with AI-based systems that can handle obscured, stained, or distant license plates more effectively

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The patent introduces an intermediary preprocessing module that enhances image quality before recognition. This module includes operations such as denoising, deblurring, and color correction that act as intermediaries between the raw image and the recognition system, improving the input quality for subsequent deep learning processing

Inventive Principle:
Principle #24Intermediary (Mediator)

2Productivity

If artificial recognition is used for vehicle identification, then the system can handle complex cases, but recognition efficiency is low and error rates are high

Engineering Contradiction:
Improverecognition efficiencyVSAvoidrecognition accuracy
Core Design Contradiction:
ProductivityVSReliability

Solution Approach 1:

The patent implements a self-service recognition system where the deep learning model automatically processes and recognizes vehicle license plates without human intervention. The system performs self-correction through the RNN's sequence modeling capabilities, which can identify and correct errors in recognition sequences, achieving both high efficiency and reliability

Inventive Principle:
Principle #25Self-service

Solution Approach 2:

The patent incorporates feedback mechanisms in the form of loss functions that guide the neural network learning process. The system uses supervised learning with labeled data, where the output is compared against ground truth and feedback is provided through gradient descent to continuously improve recognition accuracy and efficiency

Inventive Principle:
Principle #23Feedback

3Measurement precision

If deep learning models are used for vehicle recognition, then recognition accuracy improves, but computational complexity and processing time increase

Engineering Contradiction:
Improverecognition accuracyVSAvoidprocessing time
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The patent segments the recognition process into distinct stages: image preprocessing, feature extraction using CNN, and sequence recognition using RNN. This segmentation allows each module to be optimized independently, with the CNN extracting spatial features efficiently and the RNN processing temporal sequences, reducing overall processing time while maintaining high accuracy

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent performs preliminary actions through the preprocessing module that enhances image quality before the main recognition process. By applying denoising, deblurring, and color correction in advance, the system prepares optimized input data for the deep learning models, reducing the computational burden during recognition and decreasing processing time

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentEP3432214B1Method and system for vehicle recognition
Publication Date: 2024.04.17 BEIJING ICETECH SCI & TECH CO LTD
  • EP3432214B1 patent drawingFigure 1~2
  • EP3432214B1 patent drawingFigure 3
  • EP3432214B1 patent drawingFigure 4~6

AI summary

The present disclosure relates to a vehicle recognition method and system. The vehicle recognition method comprises: receiving a first image; extracting a first vehicle brand recognition region and a first color recognition region of the first image; recognizing the extracted first vehicle brand recognition region by a trained vehicle brand deep learning model; recognizing the extracted first color recognition region; and recognizing the vehicle based on the recognition results of the first vehicle brand recognition region and the first color recognition region.