Yeald-AI

home page for Yeild AI

Inspiration

This project is very personal to me. My father is a farmer, and we have cattle and buffalo at home. Growing up, I saw the challenges farmers face in managing livestock, especially with breed identification and making the right breeding choices. Many times, farmers guess or rely on local word-of-mouth to identify breeds. This often leads to lower milk yields, reduced income, and missed chances for better breeding strategies. Watching this problem within my own family made me wonder: What if technology could help? If an app could quickly identify a cow or buffalo breed from a picture and offer guidance, it would save time, reduce mistakes, and give farmers more confidence in their decisions. This idea led to the creation of this platform.

How We Built It

We gathered and organized data sets of 15 Indian cow breeds and 15 buffalo breeds since these are the most common in rural India. Using transfer learning with CNN models (ResNet/EfficientNet), we trained an image classification model to identify breeds with confidence scores. We created the backend using FastAPI/Flask and the frontend with React.js and Tailwind CSS, making the platform mobile-friendly for rural farmers.

The system not only identifies breeds but also provides:

Breed Information: origin, traits, milk yield, care tips.
Smart Recommendations: best mating companions for productivity.
Nearby Breeders: using Google Maps API (limited to India).
AI Chatbot Support: instant Q&A for livestock guidance. A feedback system allows farmers to report incorrect predictions, which helps us continuously improve the model.

What We Learned

Through this project, I learned that technology needs to fit real-world conditions. Farmers need solutions that are:

Simple to use (upload an image, get results).
Accurate in rural conditions (different lighting, animal postures).
Useful beyond identification (recommendations, breeder connections, chatbot). Most importantly, I learned how personal experiences can inspire strong innovations.

Challenges We Faced

Data Scarcity: There are very few open datasets of Indian breeds, so we had to curate and preprocess the data carefully. Accuracy in Real Conditions: The model needed to work even with imperfect photos taken on basic smartphones. Accessibility: Designing an interface that my father, who has limited digital skills, could use easily. Scalability: Balancing cost and performance for a system that could be used by thousands of farmers.

Final Impact

This project is more than just a technical prototype; it’s a solution that comes from real-life challenges on my own farm. It represents a step toward empowering Indian farmers with AI, preserving indigenous breeds, and ensuring sustainable livestock management. By turning a personal struggle into an AI-driven solution, the platform shows how innovation can emerge from the heart of rural India.