Thermodynamic Chart Conversion for Dense Crowd Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for predicting the number of people in dense crowds, such as those in public areas like retail and airports, suffer from significant errors, leading to inefficient management due to the difficulty in accurately determining crowd density and the performance degradation in high-density scenarios.
Innovation Solution
A method involving the conversion of images into thermodynamic charts using a trained thermodynamic chart conversion model, where the model is obtained by marking and training on pre-marked images, allowing for more accurate prediction of crowd density by ignoring specific areas with high density thresholds.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If detection method based on head and shoulders is used, then the number of people can be obtained according to detection boxes, but large errors occur in dense crowd scenarios
Solution Approach 1:
The patent transforms the image data into a thermodynamic chart representation, changing the parameter space from raw pixel values to thermodynamic features (temperature, pressure, volume analogs). This parameter transformation enables the model to capture crowd density patterns that are not apparent in original images, thereby improving prediction accuracy for dense crowds while maintaining adaptability across different crowd densities.
Solution Approach 2:
The patent introduces a thermodynamic chart conversion model as an intermediary between the input image and the crowd number prediction. This intermediary transforms the complex image data into a simplified thermodynamic representation that highlights crowd density characteristics, enabling more accurate predictions in dense scenarios without sacrificing the method's versatility.
2Productivity
If direct regression prediction is used, then the number of people can be obtained directly through CNN, but large errors occur in dense crowd prediction
Solution Approach 1:
The patent inserts a thermodynamic chart conversion model as an intermediary processing step between the input image and the final prediction. This intermediary transforms the image into a thermodynamic representation that preserves crowd density information more effectively, allowing the subsequent prediction model to achieve higher accuracy without significantly increasing computational complexity.
Solution Approach 2:
By converting images to thermodynamic charts, the patent changes the feature representation parameters from standard CNN features to thermodynamic analogs (temperature, pressure, volume). This parameter change enables the model to better distinguish crowd density patterns, improving prediction accuracy while maintaining computational efficiency through the structured thermodynamic feature space.
3Quantity of substance
If all areas of the image are used for training, then more data is available, but high density areas with threshold exceedance reduce prediction accuracy
Solution Approach 1:
The patent extracts and excludes high-density areas from the training data where the density threshold is exceeded. By removing these problematic regions from the training set, the model learns from reliable data regions and avoids being trained on ambiguous or erroneous high-density patterns, thereby improving overall prediction accuracy while still utilizing sufficient training data from other regions.
Solution Approach 2:
The patent applies different treatment to different regions of the image based on their density characteristics. High-density areas exceeding the threshold are excluded from training, while other regions are used normally. This local quality approach ensures that the training data quality is optimized by excluding only the problematic regions rather than reducing the overall training data volume significantly.
Data Source
AI summary
A method, apparatus, device, and storage medium for predicting the number of people of a dense crowd, including: converting a first image, in which the number of people is to be determined, into a corresponding first thermodynamic chart according to a thermodynamic chart conversion model; and determining the number of people in the first image according to the first thermodynamic chart, wherein the thermodynamic chart conversion model is obtained by training according to a pre-marked second image and a thermodynamic chart corresponding to each second image, thereby achieving prediction of the number of people of a dense crowd, improving the accuracy in predicting the number of people of the dense crowd while improving management efficiency.