A Practical Animal Detection and Collision

Avoidance System Using Computer Vision

Technique
A.Hazarathaiah1 M.Likhith Varma2 P.uday Srnivas3

1 Professor, Department of Electronics and Communication, narayana engineering college, Gudur India
2 UG student, Department of Electronics and Communication, narayana engineering college, Gudur India
3 UG student, Department of Electronics and Communication, narayana engineering college, Gudur India

ABSTRACT One serious problem that all the developed nations are facing today is death and injuries due to
road accidents. The collision of an animal with the vehicle on the highway is one such big issue, which leads
to such road accidents. In this paper, a simple and a low-cost approach for automatic animal detection on
highways for preventing animal-vehicle collision using computer vision techniques are proposed. A method
for finding the distance of the animal in real-world units from the camera mounted vehicle is also proposed.
The proposed system is trained on more than 2200 images consisting of positive and negatives images and
tested on various video clips of animals on highways with varying vehicle speed. As per the two-second rule,
our proposed method can alert the driver when the vehicle speed is up to 35 km/h. beyond this speed, though
the animal gets detected correctly, the driver does not get enough time to prevent a collision. An overall
accuracy of almost 82.5% is achieved regarding detection using our proposed method.

I. INTRODUCTION Road accidents are increasing due to the increase in a numb-

Today’s automobile design primarily depends on safety mea- beer of vehicles day by day and also the due to the absence
sores, security tools and comfort mechanism. The approach of any intelligent highway safety and alert system. According
has facilitated the development of several intelligent vehicles to data given in a study , the number of people who lost
that rely on modern tools and technology for their perform- their lives in India due to road accidents was almost 0.11
mince. The safety of an automobile is the highest priority million deaths in 2006, which was approximately 10% of the
according to a recent report . The report commissioned by total road accident deaths in the world.
World Health Organization in its Global Status Study on According to the accident research study conducted by JP
Road Safety 2013, revealed that the leading cause of death Research India Pvt. Ltd. for the Ahmedabad-Gandhi agar
for young people (15-29 age) globally is due to road traffic region (cities of India), for the duration February 2014 to
collisions. Even though various countries have initiated and January 2015, total 206 road traffic accidents were recorded
taken steps to reduce road traffic collisions and accidents, the and these were influenced by three main factors i.e. human,
total number of crashes and traffic accidents remain as high as vehicle, infrastructure or a combination of them .
1.24 million Per year . Road traffic accidents and injuries The number in figure 1 is a percentage of the total number
are expected to rise by almost 65% by the end of 2020. Due of accidents surveyed. According to the record, human factor
to road accidents, every year 1 out of 20,000 persons lose influence on road traffic accidents was 92%, vehicle 9% and
their life and 12 out of 70,000 individuals face serious injuries infrastructure 45%. Out of total 45% (91 accidents) infers-
in India . India is also known for the maximum number of structure influenced traffic accidents, 6% (12 accidents) were
road accidents in the world. According to the data given by due to animals on the road whereas out of total 92% (171)
National Crime Records Bureau (NCRB), India, there was human factor influenced traffic accidents, 14% (24) were due
almost 118,239 people who lost their life due to road to driver inattention and absence of any timely alert system
accidents in the year 2008. A major percentage of these for preventing the collision . Similar types of surveys were
road crashes and accidents involved car and other vehicles. conducted for the Mumbai-Pune expressway,
 FIGURE 1. Influences on road traffic accidents [8].

the object of interest may be falsely detected as an animal.

By JP Research India Pvt. Ltd. and the conclusions hinted at Researchers in [17] tried to discover an animal’s presence
a significant percentage of road accidents resulting due to an in the scene (image) affecting the power spectrum of the
object (animal) on the road, driver inattention, and absence of picture. This method of animal detection was also consider-
an intelligent highway safety alert system. erred not appropriate since quicker results with this approach
would involve massive amount of image processing in a short
II. LITERATURE SURVEY period [18]. Researchers in [19] also used the face detector
Applications built on detection of animals play a very vital technique initiated by Viola and Jones for a particular any-
role in providing solutions to various real-life problems. mal type. After the animal face is identified, the researchers
The base for most of the applications is the detection of track it over time. The problem with this technique is that
animals in the video or image. face detection requires animals to see into the camera not
A recent study shown that human beings have to take the necessarily captured by the road travel video. Animals can
final call while driving whether they can control their car to arrive from a scene from various directions and in different
prevent collision with a response time of 150ms or no. The sizes, poses, and colors. Another method for animal detection
issue with the above approach is that human eyes get and tracking that uses texture descriptor based on SIFT and
exhausted quickly and need rest, which is why this method matching it against a predefined library of animal textures is
is not that effective. Some scientific researchers have proposed in [20]. The problem with this method is that it is
proposed a method that requires the animals to take a pose restricted to videos having single animal only and very
towards the camera for the trigger, including face detection. minimal background clutter.
The problem with this technique is that face detection requires In Saudi Arabia, the number of collisions between the
animals to see into the camera which is, not necessarily camel and a vehicle was estimated to reach more than a Hun-
captured by the road travel video. Animals can arrive from a Dred each year [21]. Authors in [21] implemented a deploy-
scene from various directions and in different sizes, poses, able Camel-Vehicle Accident Avoidance System (CVAAS)
and color. and exploited two technologies GPS and GPRS
Animals can be detected using the knowledge of their
motion. The fundamental assumption here is that the default III. RESEARCH GAP AND CHALLENGES
location is static and can simply be subtracted. • Though various practical solutions for automatic lane
detection and pedestrian detection on highways are
All blobs, which stay after the operation are measured as the available still research related to automatic animal detect-
region of interest. Although this technique performs well in ton on highways is going on.
controlled areas, e.g. underwater videos, it does not work • Animal detection in wildlife (forest) videos or underway-
universally, especially road or highway side videos. term videos (controlled areas) have been tried in past but
Researchers used threshold segmentation approach for the challenges are much more when detecting animals on
getting the targeted animal’s details from the background. highways (uncontrolled areas) as both animal as well as a
Recent researches also revealed that it’s hard to decide the camera mounted vehicle is moving apart from other
threshold value as the background changes often. A obstacles on the road which are also moving or station-
method applicable to moving backgrounds (e.g., due to ray. There is no issue of speed (vehicle speed as well as
camera motion) is presented in subsequent studies. The animal speed) and detecting distance of animal from the
authors also state that other moving objects apart from vehicle in wildlife videos which is crucial and critical in
 animal detection on highways. color values. After color normalization, the second step of
• The biggest challenge in detecting animals compared to calculation is the computation of the gradient values. The
pedestrians or other objects is that animals come in most common method is to apply the 1D centered point
various size, shape, pose, color and their behavior is also discrete derivative mask in both the horizontal and vertical
not entirely predictable. Though the basic shape and size directions. Specifically, this method requires filtering the grey
of a human being are pretty average and standard, the scale image with the following filter kernels
same is not true for animals. So, given an image I, we obtain the x and y derivatives using
• Although various methods and approaches have been a convolution operation: IX = I∗ DX and IY = I∗ DY.
used and are still in progress to detect, solve and reduce
the number of animal-vehicle collisions, the absence of
any practical systems related to an animal-vehicle
collision on highways has delayed any substantial devil-
open in the scenario
IV. SPECIFIC REASONS FOR ANIMAL (COW)
DETECTION According to the surveys and report given by
the Society for Prevention of Cruelty to Animals (SPCA)
and [27]–[31],the number of accidents on Indian roads has
increased due to increase in a number of vehicles day by
day and also due to the presence of animals on the road
(mainly two animal’s dog and cow). The collision of an
animal with the vehicle on the highway is one such big
issue apart from other problems such as over speed, abrupt
lane change, and drunk-drive and others which lead to such
road accidents and injuries. The associated number of
fatalities and injuries are substantial too.
Specific reasons behind developing automatic cow
detection system in place of any other animal are:
• India is mainly an agriculture based country where 70%
of people depend on agriculture, and 98% of them
depend on cow based agriculture.
• The cow is a sacred animal in India and nobody wants
to hit a cow.
• Cow milk is the most useful and compatible with human
mother’s milk than any other animal or so.
• According to some surveys, cow’s milk and cow dung
have many medicinal benefits.
• Cows, as well as dogs, are found quite often than other
animals on the Indian roads.
• As cow is a large (heavy) sized animal, the collision
between a cow and vehicle will be very much severe.
The collision between a small (less weight) sized animal
like dog and car won’t be that much severe.
The speed with which the vehicle is coming and hitting the
animal also plays a critical role in deciding the impact of the
collision.

V. BRIEF OVERVIEW AND

ADVANTAGES OF HOG AND CASCADE
CLASSIFIER
A histogram of oriented gradients (HOG) is used in com-
putter vision applications for detecting objects in a video
or image, which by definition is a feature descriptor [32].
Figure 6(a) and 6(b) shows the block diagram and block
normalization scheme of HOG.
As shown in figure 6(a), first the input image is given to
color normalization block. Color normalization is used for
object recognition on color images when it is important to
remove all intensity values from the picture while preserving
Different body movement of animals to be overlooked if they
maintain a roughly upright position.
Cascading is a concatenation of various classifiers (group
based learning). The technique involves taking all the data
collected from the output of the first classifier as a supple-
monetary data for the next classifier in the group [33]. The
key advantages of boosted cascade classifiers over mono-
lithic classifiers are that it is a fast learner and requires low
computation time. Cascading also eliminates candidates
(false positives) early on, so later stages don’t bother about
them.
As shown in figure, each filter rejects non-object win-
downs and let object windows pass to the next layer of the
cascade. A window is considered as an object if and only of
all layers of the cascade classifies it as object [33]. The filter
I of the cascade is designed to
• Reject the possibly large number of non-object windows
• To allow possible large number of object windows for
quick evaluation
 pixels. Table 2 represents the relation between pixels and
VI. DISTANCE CALCULATION OF THE DETECTED meters. Graph of depth in meters versus depth in pixels was
ANIMAL As shown in figure 10, the video is taken and
plotted in Excel (figure 12) and the best fitting second order
converted into frames (ima∗ ge of size 640 480). Following is polynomial equation is
the procedure for calculating the distance of the detected
animal from the camera-mounted vehicle: y = 0.0323x2 + 22.208x + 1.3132(1)
• Image resolution is 640 × 480 Where y is the depth in pixels and x is depth in meters. In
• X range is 0 to 640 equation (2), (3) and (4), we get sensitivity close to 80.4%,
• Y range is 0 to 480 specificity close to 83.5% and accuracy of the classifier close
to 82.5%.
Let the right bottom coordinate of the detected cow be (x, y).
Figure 14 shows the on-board camera with the processing
Then the distance of cow from the lower edge (car/camera) is
480-y.

and display system inside the car on the dashboard side. We

performed extensive experiments and spent so many hours
testing the system in different weather conditions on

The road. Figure 15 shows the true positive scenario wherein

in the video, animal (cow) is present and our proposed system
correctly detects it and gives an indication (box). Similarly,
figure 16 shows a false positive case wherein animal (cow)
is detected in the video by the system even though it is absent
in that particular frame at that given location. Figure 17 shows
a false negative case wherein though the animal (cow) is
present in the video; the system indicates absence (no box)
of the animal. Figure 18 shows animal detected in the
morning condition with the experimental camera mounted
Note: The above method of distance calculation works vehicle stationary i.e. at 0 mph speed. Figure 19 shows
well with the flat ground surface. Suffers a bit if the ground animal detected in the afternoon condition with the vehicle
surface is not perfectly flat. speed at 40 mph. Figure 20 shows animal detected in the
evening state at a distance of 11 meters from the camera
VII. CONVERSION FROM PIXELS TO METERS mounted testing vehicle with the vehicle moving at a speed
There is some relationship between the depth of the object in of 60 mph. Figure 21 shows multiple animals detected in one
pixel and depth in real world units (meters) from the camera of the testing videos at a distance of 17 meters.
mounted vehicle once the object (animal) gets detected in the
frame. As the depth of the object in meters from the camera
mounted vehicle increases (size of the object decreases), the
depth in pixels also increases and vice versa [34]. This hinted
us to find a relationship between the depth of the object in
pixels and meters. Once the camera position in the car and
height of the camera from the ground was fixed (camera
calibration done), we took different images of the same object
kept at various depths from the camera center (figure 11).
The depth of the object from the camera center in meters was
known to us.
We then noted the corresponding depth of the object in
 TABLE 4. Speed-distance relation.

FIGURE 19. Animal detection at 40 mph (afternoon state).

Time and mechanical response time. As per the ‘‘two-

second rule’’ which is usually a rule of thumb suggests
that a driver should ideally stay at least two seconds
behind any object that is in front of the driver’s vehicle
[40]. The two-second rule is useful as it can be applied to
FIGURE 20. Animal detected at a distance of approximately 11 meters any speed and provides a simple and common-sense way
from the camera mounted vehicle with the speed of 60 mph in evening
time. of improving road safety. So if we go with ‘‘two-second
rule’’, clearly from Table 4 (speed-distance relation as
well as actual time (onboard) available for the driver to
responds), it indicates that when the speed of the vehicle
is between 30 to 35 mph, the driver gets some time to
apply brakes and can avoid a collision. Anything above
this speed, though the alert signal is available the driver
won’t be able to avoid a collision.

CONCLUSION
An efficient automatic animal detection and a warning system
can help drivers in reducing the number of collisions occur-
ring between the animal and the vehicle on roads and high-
ways. In this paper, we discussed the necessity of automatic
FIGURE 21. Multiple animals detected in one of the testing video animal detection system and our algorithm for animal detect-
(detecting distance of 17 meters). ton based on HOG and cascade classifier. The algorithm can
detect an animal in different conditions on highways. The
proposed system achieves an accuracy of almost 82.5 %
The camera mounted vehicle. Training and testing on large regarding animal (cow) detection. Estimation of approximate
datasets will improve the detection rate and accuracy of the animal distance from the testing vehicle is also done. Though
classifier. the proposed work has been focused on automatic animal
The average processing (computation) time with our pro- detection in context to Indian highways, it will work in other
posed image processing method is 100ms (10 frames per
second) which can be still be shortened using NVidia’s CUDA
processor. According to the article [39], the term response
time or brain reaction time of the drivers in traffic engineering
literature is composed of mental processing time, movement
 2015, JP Res. India Put Ltd., Mar. 2015.
countries also. The proposed method can easily be [9]S. Sharma and D. J. Shah, ‘‘A brief overview on different
extended for detection of other animals too after animal detect- ton method,’’ Signal Image Process., Int. J.,
proper training and testing. The proposed system vol. 4, no. 3, pp. 77–81, Jun. 2013.
[10]M. Fabre-Thorpe, A. Delorme, C. Merlot, and S. Thorpe,
can be used with other avail- able, efficient ‘‘A limit to the speed of processing in ultra-rapid visual
pedestrian and vehicle detection systems and can categorization of novel natural scenes,’’ J. Cog nit.
be offered as a complete solution (package) for Neurosis. vol. 13, no. 2, pp. 171–180, Mar. 2001.
preventing collisions and loss of human life on [11]T. Burkhart and J. Calico, ‘‘Analyzing animal behavior in
wildlife videos using face detection and tracking,’’ IEE
highways. Proc.-Vis., Image Signal Process., vol. 153, no. 8, pp.
305–312, Jun. 2006.
VIII. LIMITATIONS AND FUTURE SCOPE [12]D. Walther, D. Edging ton, and C. Koch, ‘‘Detection and
tracking of objects in underwater video,’’ in Proc. IEEE
Though our proposed system can detect the Compute. Soc. Conf. Compute. Vis. Pattern Recognition.
animals (cow) on roads and highways as well as (CVPR), Washington, DC, USA, Jun. /Jul. 2004, pp. 544–
gives alert to the driver, it has some limitations 549.
[13]J. C. Nescient and J. S. Marques, ‘‘Performance
too. The proposed system can detect animal up to evaluation of object detection algorithms for video
a distance of 20 meters only when a vehicle is surveillance,’’ IEEE Trans.
stationary. The system can prevent collision of the Multimedia, vol. 8, no. 4, pp. 761–774, Aug. 2006.
[14]N. Kawasaki, ‘‘Parametric study of thermal and chemical
veil- clef with the animal when driving at a non equilibrium nozzle flow,’’ M.S. thesis, Dept. Electron.
speed in between Eng., Osaka Univ., Osaka, Japan, 1993.
30 to 35 mph. beyond this speed, though animal [15]S. L. Hanuman, N. W. Campbell, and D. P. Gibson,
gets detected time is not sufficient to prevent ‘‘Identifying quadruped gait in wildlife video,’’ in Proc.
IEEE Int. Conf. Image Pro-
animal-vehicle collision.
Some means or method of increasing the detecting
distance of the animal from the camera mounted
vehicle needs to be done so that driver gets
sufficient time for applying brakes or take any other
action for preventing the collision which may be
solved using high-end resolution cameras or radar.
No effort has been made to detect animals during
the night, which is expected to be done in our future
scope of study and research

REFERENCES
[1] NHTSA 2020 Report, accessed on Sep. 8, 2015. [Online].
Available:
http://www.nhtsa.gov/nhtsa/whatis/planning/2020Report/
2020report.html
[2] Global Status Report on Road Safety 2013. Executive
Summary, World Health Org., Geneva, Switzerland, Oct.
2013.
[3]C. J. L. Murray and A. D. Lopez, ‘‘Alternative projections of
mortality and disability by cause 1990–2020: Global burden
of disease study,’’ Lancet, vol. 349, pp. 1498–1504, May
1997.
[4] Ministry of Home Affairs. Accidental Deaths & Suicides
in India 2012, Nat. Crime Records Bureau, New Delhi,
India, Jun. 2013.
[5]National Highways Authority of India. Information about
Indian Road Network, accessed on Jun. 3, 2010. [Online].
Available: http://www.nhai.org/roadnetwork.htm
[6]J. Padmanaban, R. Raja Raman, G. Starter, S. Narayan, and
B. Ramesh, ‘‘Analysis of in-depth crash data on Indian
national highways and impact of road design on crashes
and injury severity,’’ in Proc. 4th Int. Conf. ESAR
‘Expert Sump. Accident Res.’, Hanover, Germany, Sep.
2010, pp. 170–183.
[7] Ministry of Home Affairs. Accidental Deaths & Suicides
in India 2006, Nat. Crime Records Bureau, New Delhi,
India, Jun. 2007.
[8] Accident Research Study on the Ahmedabad–Gandhi agar
Highway for the Duration February 2014 to January