People Flow Estimation Using Particle Filter Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing people flow estimation devices struggle to accurately estimate the flow of people in crowded environments due to overlapping individuals and frequent occlusions, leading to difficulties in detecting and tracking person movements.
Innovation Solution
A people flow estimation device that includes number-of-people estimation, approximate position determination, observation model generation, and states-of-people estimation using particle filters to correct person counts, determine approximate positions, and update particle weights based on observation models and correction coefficients, allowing for accurate tracking even in crowded conditions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional upper body detection is used in crowded environments, then detection speed is maintained, but detection precision deteriorates due to overlapping and occlusion
Solution Approach 1:
The patent segments the tracking problem into two independent parts: (1) upper body detection for initial identification, and (2) head detection for continuous tracking. This segmentation allows each component to be optimized independently - upper body detection handles occlusion robustly while head detection provides precise location tracking, resolving the contradiction between detection precision and tracking complexity
Solution Approach 2:
The patent introduces an intermediary association mechanism that links upper body detection results with head detection results. The upper body detection serves as an intermediary that provides initial person identification and tracking context, which then guides the head detection process. This intermediary relationship enables the system to maintain high detection precision while avoiding the complexity of direct head detection in crowded scenes
2Measurement precision
If particle filtering is applied to improve tracking accuracy, then tracking precision improves, but computational complexity increases
Solution Approach 1:
The patent applies preliminary action by performing upper body detection first to establish initial person associations and tracking hypotheses before applying particle filtering. This preliminary step reduces the search space for the particle filter, allowing it to focus computational resources on refining tracking accuracy rather than exploring all possible associations, thus improving tracking accuracy while controlling computational load
Solution Approach 2:
The patent implements partial action by applying particle filtering selectively - using it for head position refinement and association confirmation rather than for all aspects of tracking. The upper body detection handles the bulk of tracking tasks, while particle filtering is applied partially to enhance accuracy in critical scenarios, achieving good tracking accuracy without the full computational overhead of applying particle filtering universally
Data Source
AI summary
An approximate positions-of-people determination means determines, for each partial area, approximate positions of people on an image on the basis of information about the partial area. A states-of-people estimation means, in addition to predicting the current states of particles from the states of particles indicating past states of people and stored in a storage means, adds new particles and evaluates the likelihood of the states of predicted particles and the added particles on the basis of an observation model generated for the observed people to update the weights of the particles, performs particle re-sampling at a probability proportional to the weights of those particles, and outputs the states of the obtained particles as a flow of people.