Face-Based People Counting With Short- and Long-Term Memory
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing people-counting systems using CCTV cameras face challenges with duplicate counting of individuals and high installation costs due to the need for additional devices, making it difficult to implement accurate counting without additional hardware.
Innovation Solution
A face detection system utilizing short-term and long-term memory mechanisms to track and manage detected faces, minimizing duplicate counting by registering faces in short-term memory for a preset time and transferring them to long-term memory based on matching rates and FIFO management, ensuring accurate counting without additional hardware.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If additional devices such as TOF cameras or RFID sensors are installed to improve counting accuracy, then measurement precision is improved, but device complexity and installation cost increase
Solution Approach 1:
The existing CCTV camera is made to serve multiple functions: it continues its primary surveillance function while simultaneously performing face detection and people counting. The system extracts facial features from regular CCTV video feeds without requiring dedicated counting hardware, thus achieving multi-functionality from a single device.
Solution Approach 2:
Instead of using physical additional devices like TOF cameras or RFID sensors, the system creates a virtual copy of depth/sensing information by analyzing facial features in 2D CCTV images. Face detection algorithms extract identifying characteristics that serve as a substitute for physical sensing devices, achieving the same counting objective without additional hardware.
2Measurement precision
If face detection is used to improve counting accuracy, then measurement precision is improved, but duplicate counting of the same person occurs
Solution Approach 1:
The system implements a feedback mechanism where detected faces are continuously compared against previously detected and counted faces. When a face is detected, the system checks whether it matches any already-counted individual, and only increments the counter if it's a new face. This feedback loop prevents duplicate counting while maintaining accurate tracking of unique individuals.
Solution Approach 2:
The system performs preliminary face registration and comparison before final counting. Faces are detected, extracted, and compared against the database of already-counted faces in advance, ensuring that only unique individuals are counted. This preliminary action of face matching prevents duplicate counting before it occurs.
3Productivity
If face detection algorithms are applied to CCTV video to count people, then productivity is improved, but processing time increases due to frame-by-frame analysis
Solution Approach 1:
Instead of analyzing every single video frame continuously, the system employs periodic action by detecting faces at selected intervals or triggering detection only when motion is detected. This approach maintains counting accuracy while significantly reducing the total number of frames processed, thereby decreasing processing time and improving efficiency.
Solution Approach 2:
The system performs preliminary processing by pre-extracting and storing facial features from video frames before the actual counting decision is made. This preliminary extraction allows for faster comparison and decision-making during the counting phase, reducing overall processing time while maintaining accuracy.
Data Source
AI summary
Disclosed herein is an apparatus and method for counting people based on face detection. The apparatus detects the face of a person in a video input through a camera, retrieves the detected face to check whether the detected face is a face registered in any one of short-term memory and long-term memory, counts the person of the detected face when the detected face is not retrieved from the short-term memory or the longer-term memory, registers the face of the counted person in the short-term memory, transfers the face registered in the short-term memory to the long-term memory to be registered therein when the face registered in the short-term memory remains for a preset time or longer, and deletes faces previously registered in the long-term memory in a First-In-First-Out (FIFO) manner when the number of faces registered in the long-term memory exceeds a predefined number.