SAFAR: A Stray Animal Feeding AI Reward System

Abstract

The current paper investigates how stray animals inurbanIndiafacechallengeseachdaytoobtainenoughfood. Rather than a lack of empathy from humans, the problem stems from a failure to create a systematic method for transforming empathy into practical actions. In response to this gap, the SAFAR (Stray Animal Feeding Artificial Intelligence Reward) application was created. SAFAR is a mobiletoolthatusesartificialintelligenceandgamificationto ensure that urban stray animals are fed. In terms of implementation, the technology functions in thefollowingway:ausertakesapictureusingaReactNative application, which is then immediately processed by a YOLOv8computervisionmodelonaPythonFastAPIserver. If the AI model detects both an animal and food in the photograph, the user earns credits stored in their Firebase Firestore account. Cooldown intervals are embedded to reduce the risk of cheating. It was found experimentally that image analysis took less than two seconds and worked effectively in daylight conditions. Overall, theapplication does morethanfeedstrayanimals;itcreatesadatasetthatwillbe useful for local authorities and NGOs in mapping stray animals for vaccination programs

Introduction

SAFAR (Stray Animal Feeding AI Reward) is an AI-powered mobile application designed to encourage and verify the feeding of stray animals through a reward-based system. The platform addresses the lack of organized support, recognition, and tracking for individuals who regularly feed stray animals. While many people voluntarily feed stray dogs, cats, and cattle, these efforts are often inconsistent and undocumented, making it difficult for NGOs and local authorities to monitor animal welfare activities.

The system combines computer vision and gamification to create a trustworthy feeding verification mechanism. Using a YOLOv8 object detection model, SAFAR analyzes uploaded images to confirm the presence of both a stray animal (such as a dog, cow, or cat) and feeding-related objects (such as a bowl, biscuit, roti, or hand). Once a feeding event is successfully verified, users are rewarded with points that can later be redeemed through partner organizations, encouraging sustained participation.

The platform follows a three-tier architecture consisting of a React Native mobile application, a Python FastAPI backend, and Google Firebase Firestore for real-time data storage and user management. The frontend enables user authentication, image capture, and reward tracking, while the backend processes images and performs AI-based validation. A Smart Validation Algorithm ensures that rewards are granted only when both an animal and feeding activity are detected, preventing false claims.

To maintain data integrity, SAFAR includes an anti-spam mechanism that enforces a cooldown period between submissions and uses atomic database operations to prevent duplicate rewards. Testing demonstrated successful feeding verification, contextual validation for incomplete submissions, and effective spam prevention. The system achieved end-to-end verification times of less than 2 seconds under normal lighting conditions.

The project successfully demonstrates how AI can provide proof-of-work verification for community-based animal welfare activities, creating reliable data for recognition and incentive programs. Current limitations include reduced detection accuracy in low-light conditions and dependence on GPU hardware for fast processing. Future enhancements include GPS-based heat maps, leaderboards, and partnerships with NGOs and businesses to provide meaningful rewards and support long-term participation.

Conclusion

In the SAFAR model, there is a difference between the well-meaning behavior of people and their sustained and contributinginvolvementwithsocialgood.This,however,can be addressed through the deployment of a feedback-loop system. Using computer vision to verify and gamification as motivation makes it easier for people to engage in occasional acts of social good and make themselves more visible in the process. The scope of the project suggests another important ideathatisconveyedinit,whichisthattherearemultipleuses of artificial intelligence other than those that may seem obvious at first glance.

References

[1] B.F.Skinner,ScienceandHumanBehavior.New York: Macmillan, 1953. [2] J. Redmon and A. Farhadi, \"YOLOv3: An incrementalimprovement,\"arXiv:1804.02767,2018. [3] S.Nicholson,\"Auser-centeredtheoreticalframework for meaningful gamification,\" Proc. Games+Learning+Society Conf., Madison, WI, 2012. [4] ReactNativeDocumentation.[Online].Available: https://reactnative.dev/(Accessed:Jan. 2026). [5] Ultralytics YOLOv8 Documentation. [Online]. Available:https://docs.ultralytics.com/(Accessed:Jan.2026). [6] FastAPIDocumentation.[Online].Available: https://fastapi.tiangolo.com/(Accessed: Mar. 2026). [7] GoogleFirebaseDocumentation.[Online].Available: https://firebase.google.com/docs(Accessed: Mar. 2026).

Copyright

Copyright © 2026 Jeevan Singh, Mr. Krishna Garg. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.