BHARATI VIDYAPEETH’S COLLEGE OF
ENGINEERING, NEW DELHI
Generating Security Alert in Fall
Detection System
Guided by: Mrs. Shikha Rastogi Presented by:
Muskan Bhasin (25151203119)
Shivam Jain (25251203119)
Md Faisal Hoda (20551203119)
TABLE OF CONTENT
➔ ABOUT THE PROBLEM
➔ OBJECTIVE
➔ LITERATURE SURVEY
➔ METHODOLOGY
➔ MODEL EVALUATION
➔ CONCLUSION
➔ FUTURE WORK
➔ REFERENCES
About the Problem…
Falling is the most damaging event that can happen to
an elderly person whose age is above 40s. According to
WHO, falls are the second leading cause of unintentional
injury deaths worldwide with an estimated 6,84,000
deaths last year. Falling sometimes can prove to be fatal
Fig. 1 Fall Detection System
if the person is not rescued on time. Ref:https://www.researchgate.net/figure/A-fall-detecti
on-system_fig1_344308088
The main goal of this project is to provide a tool to detect and notify the
fall of a person timely so that medical assistance can be given to present
some serious injuries.
OBJECTIVE
Our research uses an ensemble model that uses algorithms like Support Vector
Machine (SVM), K-Nearest Neighbour (KNN), Decision Tree, and Deep LSTM. The
main purpose of the project is to detect the fall of the person accurately and notify the
concerned authorities to take immediate action. It requires sending a SOS to the
registered mobile number, so that the person can get the required treatment on time.
LITERATURE SURVEY
YEAR AUTHOR TITLE TECHNIQUE USED ACCURACY
2018 Diana Yacchirema Fall detection 1. Decision trees Accuracy, Sensitivity,
Jara Suárez de Puga system for elderly 2. Ensemble and Specificity above
Carlos Palau people using IoT 3. Logistic 94%.
Manuel Esteve2 and ensemble Regression
machine learning 4. Deepnets
algorithm
2016 Pranesh Vallabh Fall detection 1) KNN Accuracy, Sensitivity,
Reza Malekian using machine and Specificity above
Ning Ye learning 83%.
Dijana Capeska algorithms
Bogatinoska
YEAR AUTHOR TITLE TECHNIQUE USED ACCURACY
2019 Na Lu Deep Learning for Fall 1. Neural Network Accuracy, Sensitivity,
Yidan Wu Detection: 2. LSTM and Specificity above
Li Feng Three-Dimensional
100%.
CNN Combined With
Jinbo Song
LSTM on Video
Kinematic Data
2018 Osvaldo Riberio IoT-Based Human Fall 1) Neural networks Accuracy, Sensitivity,
Luis Gomes Detection System and Specificity above
Zita Vale 92.5%.
METHODOLOGY
SVM
KNN
Preprocessing of
Classification
data Decision Tree Result
Deep LSTM
Ensemble Model
Sending alert in the
form of SMS/ call to
the caregiver
MODEL EVALUATION
Algorithm Used Accuracy Sensitivity Specificity
Decision Tree 93.91% 92.12 91.2
KNN 81.23% 80.32 79.34
SVM 78.34% 77.23 76.45
Deep LSTM 94.35% 93.43 92.20
Proposed 99.98% 99.80 99.90
Model(Ensemble)
CONCLUSION
With the increasing number of deaths due to falls, it's become
mandatory to develop a system to detect fall detection and alert
the concerned person/authority immediately and effectively. We
used the already built fall detection system and using twilio API as
soon as the fall is detect, an automated message will be sent to
the concerned person/authority immediately.
FUTURE WORK
In the future, we plan to add an automated calling feature too so
that as soon as the fall is detected the concerned person gets
notified immediately.
REFERENCES
[1] Gwen Bergen, Mark R. Stevens & Elizabeth R. Burns (2016). Falls and Fall Injuries Among Adults Aged ≥65 Years —
United States, 65(37);993–998
[2] U.S. Census Bureau . U.S. Population Projections: 2010 to 2050. U.S. Department of Commerce; Washington, D.C.:
(Accessed April 15, 2009). 2008.
[3] Falls in Older Persons: Risk Factors and Prevention. Accessed: Oct. 14, 2021. [Online]. Available:
https://www.ncbi.nlm.nih.gov/books/NBK235613/.
[4] Espinosa R, Ponce H, Gutiérrez S, Martínez-Villaseñor L, Brieva J, & Moya-Albor E (2019). A vision-based approach
for fall detection using multiple cameras and convolutional neural networks: A case study using the UP-Fall detection
dataset. Computers in biology and medicine, 115:103520.
[5] Cai X, Li S, Liu X, & Han G (2020). Vision-based fall detection with multi-task hourglass convolutional auto-encoder.
IEEE Access, 8:44493-44502.
[6] Li H, Li C, & Ding Y (2021). Fall detection based on fused saliency maps. Multimedia Tools and Applications,
80(2):1883-1900.
[7] Yacchirema D, de Puga JS, Palau C, & Esteve M (2019). Fall detection system for elderly people using IoT and
ensemble machine learning algorithm. Personal and Ubiquitous Computing, 23(5):801-817.
[8] Nogas J, Khan SS, & Mihailidis A (2020). Deepfall: non-invasive fall detection with deep spatio-temporal
convolutional autoencoders. Journal of Healthcare Informatics Research, 4(1):50-70.
