1

Smart Wild Animal Detection And

Alert System
Jibina Babu Lakshmi Rajesh
Department of Computer Science, Department of Computer Science,
Muthoot Institute of Technology and Science, Kochi Muthoot Institute of Technology and Science, Kochi
Email: jibinababu03@gmail.com Email: lakshmirajesh403@gmail.com
Nandhitha M Srebala R Mohan
Department of Computer Science, Department of Computer Science,
Muthoot Institute of Technology and Science, Kochi Muthoot Institute of Technology and Science, Kochi
Email: nandhithakodiyattu@gmail.com Email: srebalarmohan@gmail.com

Abstract—The Smart Wildlife Detection and Alert System is offer a highly efficient and consistent means of monitoring
an innovative solution aimed at enhancing safety and mitigat- wildlife activity and notifying relevant authorities or local
ing human-wildlife conflicts through advanced technology. The communities when animals are detected nearby. Automation in
system serves two primary purposes: protecting agricultural
and residential areas and ensuring forest-zone driving safety. In wildlife detection not only minimizes the need for continuous
farms and residential zones, high-resolution cameras and YOLO human oversight but also allows for faster, more responsive
deep learning models detect wildlife intrusions, triggering alarms alerts.Leveraging a combination of sensor technology and
and sending real-time alerts via email or SMS to safeguard machine learning algorithms, the system can detect animals
crops, property, and human lives. For forest driving, strategically in real time, which significantly improves the chances of
placed cameras detect animals near roads, with a companion
app mapping camera locations and providing real-time hazard preventing dangerous encounters or unwanted intrusions. This
alerts by changing markers to red, allowing drivers to take proactive approach benefits both humans and wildlife, as it
precautions and minimize collision risks. The system integrates allows for safer interactions and a more effective allocation of
PIR sensors, GPS tracking, and a web application to monitor conservation resources to areas where they are most needed.
and map real-time locations of vehicles and animals, enhancing Beyond immediate conflict prevention, this project is mo-
visibility and response. Alerts are managed via a Node MCU
microcontroller, which activates buzzers and sends notifications tivated by the broader potential to contribute valuable in-
to relevant authorities. This comprehensive approach not only sights to wildlife research and conservation planning.Through
reduces human-wildlife conflicts and prevents accidents but also continuous monitoring and analysis of animal behaviors and
supports wildlife conservation by promoting coexistence and movement patterns, the system can generate data that sup-
timely interventions. ports a deeper understanding of species’ needs and migration
Index Terms— YOLO, Object Detection, Sensor, GPS, GSM.
patterns.These insights can then inform decisions on habitat
preservation, migration corridor planning, and specific inter-
I. I NTRODUCTION ventions for at-risk species, helping to protect biodiversity
As human activities increasingly expand into previously un- in a targeted and data-driven way. Ultimately, this project
touched natural habitats, instances of human-wildlife conflict aspires to promote a balanced coexistence between human
have become more frequent, creating serious risks for both populations and wildlife, using smart technology to create an
humans and wildlife. These conflicts can result in physical ecosystem of peaceful cohabitation and proactive conservation.
harm to individuals, loss of property, and endangerment of By advancing our ability to monitor and respond to wildlife
animal species. Traditional methods of managing such inter- activity, this system not only addresses the immediate need
actions rely heavily on manual observation and intervention, for safety but also aligns with long-term goals of sustainable
often leading to delayed responses. These methods are not only environmental stewardship.
time intensive but also limit the ability to act quickly enough
to prevent potential conflicts or to protect endangered species II. L ITERATURE S URVEY
effectively.
Furthermore, monitoring wildlife in remote or ecologically A. Fast and Efficient Visibility Restoration Technique for
sensitive regions remains a significant challenge, as these areas Single Image Dehazing and Defogging
are difficult to access regularly, and ongoing surveillance may Authors: Apurva kumari ,Subhendu K. Sahoo , AND M.
disrupt natural ecosystems. This growing problem underscores C.Chinniah
the urgent need for innovative, technology-based solutions This paper introduces an innovative visibility restoration tech-
that can make wildlife monitoring more practical, reliable,and nique for single-image dehazing and defogging, designed to
minimally invasive. balance processing speed with high-quality haze removal. The
By harnessing advanced technologies, this system aims to methodology includes atmospheric light estimation through
 2

thresholding, transmission map refinement using dark channel vanishing gradient problem. Additionally, Soft-NMS enhances
prior and gamma transformation, and scene radiance recovery. post-processing accuracy, while motion-aware tracking ensures
The approach offers fast processing, effective haze removal, object identity is maintained across frames.
and edge preservation, demonstrating strong performance in The model demonstrates several advantages, including high
real-world and synthetic image evaluations. While the method accuracy and mAP, real-time processing at 30 FPS, and
excels in most conditions, it faces challenges in handling dense improved object localization through Wise-IoU and motion-
fog, making it a valuable yet specialized solution for real-time aware tracking. However, challenges include increased com-
applications like surveillance and intelligent vehicles. putational demands from architectural enhancements, a more
The technique employs various advanced technologies: thresh- complex training process due to dataset augmentation and
olding for atmospheric light estimation, gamma transformation architectural upgrades, and potential performance issues under
for refining depth and visibility, and median filtering to smooth extreme low-light conditions or heavy occlusions.
transmission maps while retaining edge details. Dark channel To address these limitations, the project proposes reducing
prior aids initial transmission map estimation, while guided computational overhead through model pruning, simplifying
and bilateral filtering further enhance edge preservation and in- the training process using transfer learning with pre-trained
tensity smoothing. Mathematical modeling underpins the visi- models, and expanding the dataset to include diverse real-
bility restoration process, incorporating exponential functions world scenarios, such as varying lighting conditions, object
to relate atmospheric parameters. Performance metrics such as types, and motion patterns. These improvements aim to make
SSIM, PSNR, MSE, and contrast gain ensure comprehensive the model more efficient and adaptable while maintaining its
evaluation of restored image quality. strong detection and localization capabilities.
Key advantages of this technique include improved visibil-
ity in foggy or hazy images, robust edge preservation, re-
duced computational complexity compared to deep learning C. IoT Sensor Network for Wild-Animal Detection near
approaches, and high quantitative and qualitative accuracy. Roads
Its cost-effectiveness, stemming from single-image process- Authors: Mindaugas Knyva, Darius Gailius, Gintautas
ing, further enhances its suitability for real-time applications. Balčiūnas, Darius Pratašius and Pranas Kuzas
However, limitations include reduced effectiveness in dense The project focuses on a system designed to detect wild
fog, halo artifacts near edges, dependency on assumptions animals near roads using IoT technologies to prevent vehicle-
like constant atmospheric light, and the lack of adaptability animal collisions. The system operates autonomously, powered
or learning-based generalization. Additionally, manual tuning by solar panels, and includes sensor nodes equipped with
of parameters like gamma values can be time-intensive. Passive Infrared (PIR) sensors for motion detection, LoRa
To improve this technique, dynamic gamma adjustment based modules for communication, and LEDs for signaling. A ther-
on haze density and localized atmospheric light estimation for mal camera provides additional verification data, enhancing
non-uniform haze or dense fog are proposed. These enhance- the reliability of the system. Custom software integrates the
ments aim to make the method more adaptive and effective sensors, processes signals, triggers alerts, and logs data for
across varied atmospheric conditions, addressing its current analysis. The system is low-cost, energy-efficient, and easily
limitations while maintaining its efficiency and practicality for deployable, providing real-time safety alerts for drivers in
real-time applications. wildlife-prone areas.
Key technologies used in the system include PIR sensors for
energy-efficient motion detection, LoRa modules for long-
B. An Improved YOLOv8 to Detect Moving Objects range communication, and solar panels for sustainable power.
Authors: Mukaram Safaldin,Nizar Saghden,Mahmoud Mej- The thermal camera, used for validation, helps confirm the
doub presence of animals detected by PIR sensors. The gateway
This paper presents an enhanced version of the YOLOv8 collects data from sensors, and LEDs provide visual warnings
model, tailored for detecting moving objects in visual sce- to drivers. The system’s design is scalable, allowing for the
narios with high accuracy and real-time performance. The installation of nodes at regular intervals along roadsides,
framework integrates advanced preprocessing techniques, ar- making it adaptable to various lengths of roadway. Its long-
chitectural modifications, and data augmentation strategies to range communication ensures that the system remains effective
optimize detection and localization. Evaluated on benchmark in rural or forested areas, even with limited infrastructure.
datasets such as KITTI, LASIESTA, PESMOD, and MOCS, The system’s advantages include energy efficiency, cost-
the model achieves 90 effectiveness, scalability, and real-time safety alerts. However,
The proposed methodology incorporates YOLOv8 for object it faces challenges such as limited detection range of PIR
detection, preprocessing techniques like frame extraction and sensors, susceptibility to false positives from environmental
background removal, and data augmentation strategies such factors, and limitations in camera resolution and performance
as geometric transformations and synthetic motion blur. Key in low-light conditions. The system also depends on solar
innovations include Ghostblock integration to reduce compu- energy, making it vulnerable to overcast conditions or short
tational complexity, the Wise-IoU loss function for enhanced daylight hours, which can reduce efficiency. Regular main-
bounding box regression, and attention mechanisms with resid- tenance is needed to clean sensor optics and solar panels,
ual networks to improve feature extraction and mitigate the and low temperatures can affect battery performance. These
 3

limitations highlight the need for improvements in detection III. C ONCLUSION

accuracy, energy efficiency, and overall system reliability In conclusion, the Smart Wildlife Detection and Alert
under various environmental conditions. System integrates advanced technologies like YOLOv8, IoT
Research gaps identified include false triggers from environ- sensors, and GSM-GPS for effective wildlife monitoring and
mental factors, sensor contamination, and energy harvesting human-wildlife conflict prevention. By providing real-time
limitations in adverse weather. Additionally, the thermal cam- alerts and leveraging solar-powered nodes, the system en-
era’s inability to effectively verify sensor triggers in darkness hances safety in agricultural and road environments. Despite
and its low resolution poses a challenge for accurate animal challenges like network dependency and false positives, on-
detection. Addressing these gaps and improving the system’s going improvements, such as dynamic gamma adjustment
performance under challenging conditions will be crucial for and hybrid communication protocols, will optimize its per-
enhancing its effectiveness in real-world deployments. formance for real-world wildlife protection applications.

R EFERENCES
[1] Chih-Yuan Lien, Chi-Huan Tang, Pei-Yin Chen, Yao-Tsung Kuo, and
D. ]GSM-GPS-Based Animal Tracking System: Improving Yue-Ling Deng,”A Low-Cost VLSI Architecture of the Bilateral Filter
Wildlife Monitoring Efforts for Real-Time Image Denoising”,IEEE Access, vol. 8, pp.64278-64283,
Apr. 2020.
[2] Zhang, X., Li, Y., Zhang, H. (2024). ”Forest nighttime image en-
Authors: Gayatri Mohanta,Prasad L. B.,Mohanta hancement based on improved SCI light estimation and reflection
The research paper presents a GSM-GPS-based Animal Track- optimization”. IET Image Processing, 18, 1028–1041
ing System (ATS) designed to improve wildlife monitoring [3] Dave, B., Mori, M., Bathani, A., Goel, P. (2023). ”Wild Animal
Detection using YOLOv8”. Procedia Computer Science, 230, 100-111
and prevent animal intrusion into human-populated areas. The [4] R.Tapiador-Morales et al., ”System based on inertial sensors for behav-
system integrates GPS and GSM technologies to provide real- ioral monitoring of wildlife”, Computer, International Conference on
time data on animal movements, behavior, and environmen- Information andTelecommunication Systems 2015.
[5] Rahman, Z., Pu, Y.-F., Aamir, M., Wali, S.: Structure revealing of
tal conditions. It monitors factors such as heart rate, body low-light images using wavelet transform based on fractional-order
temperature, and surrounding environmental elements, with denoising and multiscale decomposition. Vis. Comput. 37, 865–880
alerts sent via SMS to authorities in case of unusual activity (2021). doi:10.1007/s00371-020-01838-0
[6] M. Ju, C. Ding, Y. J. Guo, and D. Zhang, “IDGCP: Image dehazing
or potential threats. The ATS aims to enhance conservation based on gamma correction prior,” IEEE Trans. Image Process., vol. 29,
efforts by offering valuable insights into animal migration, pp. 3104–3118, 2020
habitat usage, and population trends while preventing human- [7] Ma, L., Ma, T., Liu, R., Fan, X., Luo, Z.: Toward fast, flexible, and
robust low-light image enhancement (2022)
animal conflicts. [8] R.Tapiador-Morales et al., System based on inertial sensors for be-
Key technologies used in the system include the Global Posi- havioral monitoring of wildlife, Computer, International Conference on
tioning System (GPS) for tracking animal locations, the Global Information andTelecommunication Systems 2015.
[9] J. I. Miano et al., “Microcontroller-based vessel passenger tracker using
System for Mobile Communication (GSM) for wireless data GSM System: An aid for search and rescue operations, ” Int. J. Adv.
transmission and SMS alerts, and microcontrollers (typically Comput. Sci. Appl., vol. 10, no. 9, pp. 1–19, 2019.
Arduino-based) for data processing. Sensors track the physio-
logical parameters of the animals, and specialized software,
including libraries such as TinyGPS++ and SoftwareSerial,
is used for data integration and communication. The system
relies on serial communication protocols to transfer data and
send alerts using AT commands.
The system’s advantages include real-time monitoring, health
tracking, and minimal disruption to animals due to its
lightweight, non-intrusive design. It supports conservation ef-
forts by providing crucial data on animal behavior and habitat
usage. The system is also cost-effective, using affordable
and widely available technologies that can be adapted for
different species. However, the system has some limitations,
including dependency on GSM network availability, the risk
of equipment loss, potential data privacy concerns, and energy
consumption for powering the tracking devices. There is also a
possibility of discomfort or harm to animals due to the devices.
Improvements for the project include enhancing the tracking
system with more precise GPS data, enabling real-time moni-
toring and mapping of animal movements. Additionally, hybrid
communication protocols could be introduced, supplementing
GSM with other methods like LoRa, which can be more
effective in areas with limited GSM coverage, ensuring reliable
data transmission in remote regions.