SOS Signal Identifying Application Embedded in CCTV
Introdution:
-Reasons:
Today, human life is developing day by day, but the danger is still increasing.
Therefore, CCTV is everywhere in order to manage, ensure security, manage
traffic, ... requires and needs the support of science and technology. One of the
great supports is to help victims get support quickly and conveniently, as soon
as they give out a distress signal.
- Purpose:
detect unusual actions and signs of distress //from individuals, then
immediately send an alert to the security system or other rescue systems,
to promptly handle incidents in the area.
-meaning:
Ensuring safety //is essential to creating a healthy society and community.
Serving the protection of the people, contributing to the development of
the country.
Research Methods
The software predicts actions with signs of distress for identification (hands-on
actions, “SOS”, shouting, facial expressions, ...) and input data. When places
with cameras record and transmit data to the computer, the software will extract
the actions and sounds featured in the video, which are fed into the algorithm
and compared with the data set. After an action that matches the data
immediately sends out a signal, send an alert to the security system.
Function of SIAEC
//tiếp nhận nguồn dữ liệu thời gian thực được thu trực tiếp từ camera tích hợp ->phân
tích và nhận dạng các đặc trưng của từng đối tượng.( phân tích các cử chỉ tay, biểu
cảm khuôn mặt, hoạt động vật lý của cơ thể và âm thanh nói)
receive real-time data source directly from the built-in camera -> analyze and
recognize the characteristics of each object. (analysis of hand gestures, facial
expressions, physical activities of the body, spoken sound) -> The software will
review and calculate the percentage of the object that can signal for help -> send
a report to the observation center and solve the problem.
Service purpose of SIAEC
security, health, traffic
Technologies applied to SIAEC software:
Voice recognition(Hidden Markov Model)
Hand gesture recognition(Convolutional Neural Networks)
Detailed body movement recognition(YOLOv7 algorithm)
Working principle of SIAEC
Programming languages, libraries,...:
Pyhthon, OpenCV, Tensor Flow, Tensor Board
Steps to take:
Data collection: collect a sufficiently large number of data sets on distress
signals (hand actions, sos, screams, facial expressions, ..). Data needs to be
correctly labeled to ensure model accuracy.
Data processing : Collected data must be preprocessed before being included in
the model.
Model building: This model can be built using machine learning techniques,
deep learning, or a combination of methods.
Model training and testing: train and test the model with labeled data sets to
evaluate the accuracy of the model.
Model deployment: the model can be deployed in the surveillance system to
recognize the distress signal via CCTV.
related document
[1] B. S. Atal, M. R. Schroeder, and V. Stover, Voice-Excited Predictive
Coding System for Low Bit-Rate Transmission of Speech, Proc. ICC,
pp.30-37 to 30-40, 1975.
[2] Real-Time Hand Gesture Recognition Based on Deep Learning
YOLOv3 Model. -https://www.mdpi.com/2076-3417/11/9/4164.
[3] YOLOv7: A Deep Learning Framework for Real-Time Object
Detection - https://arxiv.org/abs/2108.07962
[4]"EgoGesture: A New Dataset and Benchmark for Egocentric Hand
Gesture Recognition" (ICCV, 2017)
[5]"NTU RGB+D: A Large Scale Dataset for 3D Human Activity
Analysis." IEEE Conference on Computer Vision and Pattern
Recognition (CVPR), 2016.
[6]Speech Commands: A Dataset for Limited-Vocabulary Speech
Recognition - https://arxiv.org/abs/1804.03209
[7]Convolutional Neural Networks for Hand Gesture Recognition
(https://www.researchgate.net/publication/326162611_Convolutional_Ne
ural_Networks_for_Hand_Gesture_Recognition)
[8]"CCTV Networking and Digital Technology," của Vlado
Damjanovski.
SOS signal Identifying
Application Embedded in CCTV
