Webology (ISSN: 1735-188X)

Volume 18, Number 6, 2021

Animal Detection And Alert System Using Computer

Vision Technique

Ponmani R

Assistant Professor, Department of Computer Science and Engineering,

Chennai Institute of Technology, Chennai.

ABSTRACT
The integration of computer vision techniques has revolutionized the field of object detection,
enabling the development of intelligent systems for various applications. This paper presents
an Animal Detection and Alert System leveraging the Speeded-Up Robust Features (SURF)
algorithm in computer vision. The proposed system aims to detect and recognize animals in
real-time by analyzing video streams or images captured through cameras. The SURF
algorithm, known for its robustness to changes in scale, rotation, and illumination, is utilized
for feature extraction, enabling the identification of distinctive patterns and keypoints within
images. These extracted features serve as the basis for training a machine learning model,
facilitating the classification and recognition of different animal species. The system is
designed to operate in diverse environmental conditions and is adaptable to various species,
accommodating a wide range of animal shapes, sizes, and appearances. Upon successful
detection, the system triggers an alert mechanism to notify users or relevant authorities,
contributing to the timely and effective management of wildlife crossings, habitat monitoring,
or animal presence in restricted areas. This research outlines the methodology employed in
dataset collection, preprocessing, feature extraction using SURF, model training, real-time
implementation, and alert mechanisms. Furthermore, it addresses the challenges encountered
in deploying such systems and proposes avenues for future enhancements, including
optimizing performance, expanding the dataset, and exploring alternative algorithms for
improved accuracy and efficiency. The proposed Animal Detection and Alert System using
omnipotent classifier which demonstrates promising capabilities in enhancing wildlife
conservation efforts, minimizing human-wildlife conflicts, and supporting ecosystem
monitoring, contributing significantly to the field of computer vision applications in wildlife
management and environmental conservation.
Keywords: Speeded-Up Robust Features, Animal Detection, Alert System, Object Detection,
Omnipotent Classifier, Scale, Rotation, and Illumination.
1 INTRODUCTION
In the prior days, the forest regions are extremely enormous. Because of deforestation, trees are
currently worked as huge structures. As the human population is as yet expanding, woodland
zones will be totally crushed. Human-Animal struggle in the general public is for the most part
because of lack of nourishment for creatures. Animal aggravation generally happens in the
woods fringe regions amongst sunset and first light and it is emphatically regular, comparing
9321 http://www.webology.org
Webology (ISSN: 1735-188X)
Volume 18, Number 6, 2021

with edit collecting periods [1]. This has been the primary driver for almost 80 human passing
and more than 300 elephant passing that have been recorded every year in the current past.
Rivalry for space is the essential purpose behind clash amongst people and Animals [2]. Wild
animals that escape from the natural life national parks wander into towns making devastation
afterward. To counteract such accidents, an appropriate framework is required to contain and
screen elephants in national parks. Here we portray diverse ways to deal with identify elephants
and conceivable methods for checking the national untamed life parks [3]. The examination,
researches and investigations the different location and following methods utilized for before
days. Be that as it may, satisfactory and proper technique has not yet grown instead of averting
strategy. In this proposed framework, HEC is diminished utilizing the vibration sensor [4]. The
vibration sensors are placed on the forest border areas. The arrival of elephant will be detected
by sensor and the flag created will be sending to the microcontroller for recognition reason.
After the compliance of elephant landing by coordinating procedure the content will be send to
the timberland authorities. Taken after by the instant message, the sound is made to alarm the
town individuals. Accordingly, our proposed framework will conquer the downsides in the
current framework by following and identifying the elephant viably [5].
In our research, Human Animal conflict is major problem in village areas like nilgris,
muthumalai, etc. Frequent co-incidence between human and wild animal causes severe damage
to farmlands, humans, as well as elephants [6]. The proposed paper is very efficient and
effectively reduces the human animal conflict. Camera is used to track the wild animal arrival
in forest borders areas. Based on sensor data’s, corresponding animal is detected through the
camera, it sends information to forest officials and the village members [7]. The animal is
matched with unique features of the image, which is stored in the database. The SURF
algorithm is used to match the captured image. Immediate alert has to send to nearby village.
2 RELATED WORKS
The aim of our proposed scheme is vission based animal detection system. In this method
camera is used to detect the animals arrival in the forest areas to prevent human elephant
conflicts. The system deters the elephant while crossing the boundary that it helps to prevent
loss of human life and damage of crops. The alarm system to alert people and intimation of
elephant arrival to forest officials will be more efficient.
RF sensor networks are wireless networks that can localize and track people (or targets) without
needing them to carry or wear any electronic device [8]. They use the change in the received
signal strength (RSS) of the links due to the movements of people to infer their locations [9].
In this paper, we consider real-time multiple target tracking with RF sensor networks. We apply
radio tomographic imaging (RTI), which generates images of the change in the propagation
field, as if they were frames of a video. Our RTI method uses RSS measurements on multiple
frequency channels on each link, combining them with a fade level-based weighted average
[10]. We introduce methods, inspired by machine vision and adapted to the peculiarities of
RTI, that enable accurate and real-time multiple target tracking. Several tests are performed in
an open environment, a one-bedroom apartment, and a cluttered office environment. The
results demonstrate that the system is capable of accurately tracking in real-time up to four
targets in cluttered indoor environments, even when their trajectories intersect multiple times,

9322 http://www.webology.org
Webology (ISSN: 1735-188X)
Volume 18, Number 6, 2021

without mis-estimating the number of targets found in the monitored area [11]. The highest
average tracking error measured in the tests is 0.45 m with two targets, 0.46 m with three
targets, and 0.55 m with four targets.
This paper presents a driving simulator experiment, which evaluates a road-departure
prevention (RDP) system in an emergency situation. Two levels of automation are evaluated:
1) haptic feedback (HF) where the RDP provides advisory steering torque such that the human
and the machine carry out the maneuver cooperatively, and 2) drive by wire (DBW) where the
RDP automatically corrects the front-wheels angle, overriding the steering-wheel input
provided by the human [12]. Thirty participants are instructed to avoid a pylon-confined area
while keeping the vehicle on the road. The results show that HF has a significant impact on the
measured steering wheel torque, but no significant effect on steering-wheel angle or vehicle
path. DBW prevents road departure and tends to reduce self-reported workload, but leads to
inadvertent human-initiated steering resulting in pylon collisions. It is concluded that a low
level of automation, in the form of HF, does not prevent road departures in an emergency
situation [13]. A high level of automation, on the other hand, is effective in preventing road
departures. However, more research may have to be done on the human response while driving
with systems that alter the relationship between steering-wheel angle and front-wheels angle.
This paper introduces two main contributions to the wireless sensor network (WSN) society.
The first one consists of modeling the relationship between the distances separating sensors
and the received signal strength indicators (RSSIs) exchanged by these sensors in an indoor
WSN [14]. In this context, two models are determined using a radio-fingerprints database and
kernel-based learning methods. The first one is a non-parametric regression model, while the
second one is a semi-parametric regression model that combines the well-known log-distance
theoretical propagation model with a non-linear fluctuation term. As for the second
contribution, it consists of tracking a moving target in the network using the estimated
RSSI/distance models [15]. The target’s position is estimated by combining acceleration
information and the estimated distances separating the target from sensors having known
positions, using either the Kalman filter or the particle filter. A fully comprehensive study of
the choice of parameters of the proposed distance models and their performances is provided,
as well as a study of the performance of the two proposed tracking methods. Comparisons to
recently proposed methods are also provided.

3 PROPOSED MODEL
The aim of our proposed scheme is vission based animal detection system. In this method
camera is used to detect the animals’ arrival in the forest areas to prevent human elephant
conflicts. The system deters the elephant while crossing the boundary that it helps to prevent
loss of human life and damage of crops. The alarm system to alert people and intimation of
elephant arrival to forest officials will be more efficient. In this system camera is used to detect
the movement of the wild animal. This camera is placed in the forest border areas. When
dangerous arrived in the forest border areas, if it crosses the threshold value the analog input is
send to the microcontroller. As per our survey and knowledge, there is no public database of
animals under consideration exists in the literature. As our research is focussed on automatic

9323 http://www.webology.org
Webology (ISSN: 1735-188X)
Volume 18, Number 6, 2021

animal (cow) detection in context to Indian roads and conditions, we didn't get much data
(images or videos) of cow needed to build a robust database. Even though a good source for
the animal images is the KTH dataset [1] and NEC dataset [2] that included pictures of cow
and other animals, some more animal videos and pictures have been clicked (during different
weather conditions i.e. morning, afternoon and evening) and some of the images have been
collected from the internet for creating a healthy database. Hence, a new animal database is
created. It is very much essential to have good database and at the same time the quality of the
database directly affects the final performance of the classifier. Already the camera is in the
ON state when the signal reaches from microcontroller to camera features start matching with
the actual image with stored image. The images are already stored in the database for finding
the animal which is entered in the forest border areas. For features matching SURF (speedup
robust future) algorithm is used to match the unique features of different animals. The images
are stored in the blob in the MySQL. Once animal is detected the emergency response is given
to the forest officials through e-mail. To alter the village members the information passed
through Zigbee protocol. The biggestissue or the problem indetecting animals compared to
pedestrians or other objects is that animals come in various size, shape, pose, color and their
behavior is also entirely unpredictable. It is entirely unrealistic to build an omnipotent
classifier, which can recognize all kind of animals with a casual pose.

Data Collection Feature omnipotent Real Time

and Pre- Extraction classifier Detection &
processing using SURF Alert System

Performance
Evaluation

Figure 1: Proposed Architecture

The block diagram consists of two modules as software and hardware module. The process
starts with the sensing camera which is to sense the arrival of elephant. when it exceeds then
the signal is send to the microcontroller. The analog signal from the microcontroller is send to
the PC by serial port communication. The camera is connected to the PC for viewing and
capturing the animal image which is then used for matching. The ZigBee transmitter and
receiver are used for short distance communication. The LCD display is used to display the
elephant name from the database. The buzzer is connected at the receiver side to provide alert
signal. The software module consists of MS-visual studio to develop computer programs and
surf algorithm for detecting the interest points on image by using blob detector. The surf
algorithm is used to locate and recognize the animal and matching its features.

SURF Algorithm

9324 http://www.webology.org
Webology (ISSN: 1735-188X)
Volume 18, Number 6, 2021

Speeded up robust features (SURF) is a patented local feature detector and descriptor. It can
be used for tasks such as object recognition, image registration, classification or 3D
reconstruction. It is partly inspired by the scale-invariant feature transform (SIFT) descriptor.
The standard version of SURF is several times faster than SIFT and claimed by its authors to
be more robust against different image transformations than SIFT. To detect interest points,
SURF uses an integer approximation of the determinant of Hessian blob detector, which can
be computed with 3 integer operations using a precomputed integral image. Its feature
descriptor is based on the sum of the Haar wavelet response around the point of interest. These
can also be computed with the aid of the integral image. SURF descriptors have been used to
locate and recognize objects, people or faces, to reconstruct 3D scenes, to track objects and to
extract points of interest. The image is transformed into coordinates, using the multi-resolution
pyramid technique, to copy the original image with Pyramidal Gaussian or Laplacian Pyramid
shape to obtain an image with the same size but with reduced bandwidth. This achieves a
special blurring effect on the original image, called Scale-Space and ensures that the points of
interest are scale invariant.
SURF uses square-shaped filters as an approximation of Gaussian smoothing. (The SIFT
approach uses cascaded filters to detect scale-invariant characteristic points, where the
difference of Gaussians (DoG) is calculated on rescaled images progressively.) Filtering the
image with a square is much faster if the integral image is used. The sum of the original image
within a rectangle can be evaluated quickly using the integral image, requiring evaluations at
the rectangle's four corners. SURF uses a blob detector based on the Hessian matrix to find
points of interest. The determinant of the Hessian matrix is used as a measure of local change
around the point and points are chosen where this determinant is maximal. The box filter of
size 9×9 is an approximation of a Gaussian with σ=1.2 and represents the lowest level (highest
spatial resolution) for blob-response maps.
The feature finding process is usually composed of 2 steps. The first step is to find the interest
points in the image which might contain meaningful structures; this is usually done by
comparing the Difference of Gaussian (DoG) in each location in the image under different
scales. A major orientation is also calculated when a point is considered a feature point. The
second step is to construct the scale invariant descriptor on each interest point found in the
previous step. To achieve rotation invariant, we align a rectangle to the major orientation. The
size of the rectangle is proportional to the scale where the interest point is detected. The
rectangle is then cropped into a 4 by 4 grid.
4 EXPERIMENTAL RESULTS
When an elephant arrive in forest border areas, vibration sensor gets the signal and the elephant
is detected. The features of the elephant are mapped with the existing data in the database. If
the two features match, an alarm gets ON with an instant e-mail to the forest officials and the
village people are alerted with the help of an LCD display regarding the elephant‟s arrival.

9325 http://www.webology.org
Webology (ISSN: 1735-188X)
Volume 18, Number 6, 2021

Figure 2: Features mapping

The above figure shows that features mapping of the animals using SURF algorithm. The image
which is stored in the database is matched with image which is present in the camera.

Figure 3: Application Programming Interface

The above figure shows the overall experimental setup of the human elephant conflict using
the vibration of the elephant. The detection is carried by the vibration sensor and the features
matching is done and emergency response given to forest officials and the village members.
5 CONCLUSION
In this paper, we focus on wildlife monitoring and analysis through animal detection from
natural scenes acquired by camera-trap networks. The image sequences obtained from camera-
trap consist of highly cluttered images that hinder the detection of animal resulting in low-
detection rates and high false discovery rates. To handle this problem, we have used a camera-
trap database that has candidate animal proposals using multilevel graph cut in the spatio-
temporal domain. These proposals are used to create a verification phase that identifies whether
a given patch is animal or background. Human Animal conflict is major problem in village
areas like nilgris, muthumalai, etc. Frequent co-incidence between human and wild animal
causes severe damage to farmlands, humans, as well as elephants. The proposed paper is very
efficient and effectively reduces the human animal conflict. Camera is used to track the wild
animal arrival in forest borders areas. Based on sensor data’s, corresponding animal is detected
through the camera, it sends information to forest officials and the village members. The animal
is matched with unique features of the image, which is stored in the database. The SURF
algorithm is used to match the captured image. Immediate alert has to send to nearby village.

9326 http://www.webology.org
Webology (ISSN: 1735-188X)
Volume 18, Number 6, 2021

In future, the automatic sound amplification device can be designed to create panic to the
elephant through huge noise. This will avoid Human Elephant Conflict in forest border area.
REFERENCES
1. Witczuk, J., Pagacz, S., Zmarz, A., & Cypel, M. (2018). Exploring the feasibility of
unmanned aerial vehicles and thermal imaging for ungulate surveys in forests-preliminary
results. International Journal of Remote Sensing, 39(15-16), 5504-5521.
2. Sureshkumar, S., Janet, J., Srivaishnavi, D., & Biruntha, S. (2020). An efficient
surveillance system to detect and prevent elephant intrusion in forest borders. Journal of
Computational and Theoretical Nanoscience, 17(4), 1725-1727.
3. Antônio, W. H., Da Silva, M., Miani, R. S., & Souza, J. R. (2019). A proposal of an animal
detection system using machine learning. Applied Artificial Intelligence, 33(13), 1093-
1106.
4. SEETHAMANI, M. P., BHARANIDHARAN, R., & LOGANATHAN, M. (2019). Forest
Animal Detection System Using Machine Learning.
5. Merkle, J. A., Monteith, K. L., Aikens, E. O., Hayes, M. M., Hersey, K. R., Middleton, A.
D., ... & Kauffman, M. J. (2016). Large herbivores surf waves of green-up during
spring. Proceedings of the Royal Society B: Biological Sciences, 283(1833), 20160456.
6. Stewart, F. E., Fisher, J. T., Burton, A. C., & Volpe, J. P. (2018). Species occurrence data
reflect the magnitude of animal movements better than the proximity of animal space
use. Ecosphere, 9(2), e02112.
7. Middleton, A. D., Merkle, J. A., McWhirter, D. E., Cook, J. G., Cook, R. C., White, P. J.,
& Kauffman, M. J. (2018). Green‐wave surfing increases fat gain in a migratory
ungulate. Oikos, 127(7), 1060-1068.
8. Dua, I., Shukla, P., & Mittal, A. (2015, December). A vision based human-elepahant
collision detection system. In 2015 Third International Conference on Image Information
Processing (ICIIP) (pp. 225-229). IEEE.
9. Andavarapu, N., & Vatsavayi, V. K. (2017). Wild-animal recognition in agriculture farms
using W-COHOG for agro-security. International Journal of Computational Intelligence
Research, 13(9), 2247-2257.
10. DeNezzo, N. (2019). Taking the Bite out of the Bight: An Assessment of Non-Lethal Shark
Bite Mitigation Strategies and Potential Applications in Southern California.
11. Jeevitha, S., & Kumar, V. (2020). A review of animal intrusion detection system. Int. J.
Eng. Res. Technol, 9(5), 1219-1221.
12. Begum, M., Janeera, D. A., & AG, A. K. (2020, February). Internet of things based wild
animal infringement identification, diversion and alert system. In 2020 International
Conference on Inventive Computation Technologies (ICICT) (pp. 801-805). IEEE.
13. Duan, Y., Li, D., Li, Z., & Fu, Z. (2015). Review on visual attributes measurement research
of aquatic animals based on computer vision. Transactions of the Chinese Society of
Agricultural Engineering, 31(15), 1-11.
14. Rey, N., Volpi, M., Joost, S., & Tuia, D. (2017). Detecting animals in African Savanna
with UAVs and the crowds. Remote Sensing of Environment, 200, 341-351.

9327 http://www.webology.org
Webology (ISSN: 1735-188X)
Volume 18, Number 6, 2021

15. Van De Gucht, T., Van Weyenberg, S., Van Nuffel, A., Lauwers, L., Vangeyte, J., & Saeys,
W. (2017). Supporting the development and adoption of automatic lameness detection
systems in dairy cattle: Effect of system cost and performance on potential market
shares. Animals, 7(10), 77.

9328 http://www.webology.org