Problem Statement: Farmers can take pictures of crops and describe problems in simple voice; the app uses CV + NLP to detect issues and give basic solutions.
Tools used: TensorFlow, OpenCV, NLP
Objective: PlantiFi is an AI-driven platform that addresses key agricultural challenges by enabling image-based plant diagnostics, providing instant health assessments, delivering detailed disease and treatment information, supporting multilingual text and audio outputs, and operating fully offline for universal accessibility.
Insights:
The tool empowers small and marginal farmers with early crop-disease detection, optimized resource use, and improved productivity. Rural communities benefit from a low-resource, internet-independent solution that encourages knowledge-sharing and reduces reliance on costly external services. Socio-economic impacts include increased crop yield and income, reduced treatment costs, empowered decision-making through expert insights, and strengthened national food security.
Problem Statement: Traditional logistics systems prioritize cost and time, often neglecting environmental impact. Develop a deep learning model that uses route data, fuel usage, weather, traffic, and cargo weight to optimize delivery routes for minimal carbon emissions, helping companies make green logistics decisions.
Tools used: Streamlit, Tensorflow,CV Pandas, Matplotlib
Objective: The core objective of this solution is to help users compare
multiple delivery routes and select the one with the lowest
predicted carbon emissions. This not only enhances
operational efficiency but also aligns with sustainability
goals.
Insights:
A smart, ML-powered Streamlit application that compares delivery routes and identifies the option with the lowest predicted carbon emissions using real logistics data. Users configure three routes with variables like vehicle type, distance, cargo weight, weather, and traffic. A TensorFlow/Keras deep-learning model predicts emissions after preprocessing (one-hot encoding, scaling). The tool enhances route planning efficiency, supports sustainability goals, and empowers businesses to make data-driven, eco‑friendly logistics decisions through an intuitive, real-time interface.
Problem Statement: In today’s fast-paced world, people are becoming more aware of the importance of healthy eating — but not everyone knows what ""healthy"" really means for them. A diet that works for one person may be harmful to another, especially for those living with medical conditions like diabetes, high blood pressure, thyroid issues, or severe allergies.
Most of the diet plans available online are generic — built for the average" person. But there’s really no such thing when it comes to food and health. What people truly need is a plan that understands their body, their medical history, their goals, and even their preferences.
This is where personalized nutrition becomes more than just a convenience — it becomes a necessity.
Tools used: Python, Flask, HTML,CSS,Gemini API, Pandas, SMTP
Objective: To bridge the gap between generic diet advice and truly personalized nutrition, we propose an AI-powered diet planning system that adapts to each individual’s unique health profile, goals, and food preferences.
By leveraging Google’s Gemini API, the system generates custom meal plans that are not only nutritionally sound but also aware of medical conditions such as diabetes, hypertension, or allergies. Users simply fill out a health profile, and the system automatically adjusts calorie goals, macronutrient ratios, and dietary restrictions to match their needs.
But it doesn’t stop there — the app also learns over time. Through built-in feedback mechanisms, users can rate their meals, and the AI fine-tunes future suggestions accordingly. This creates a smarter, more user-friendly experience that improves with each interaction.
Insights:
Provides health‑aware meal plans tailored to medical conditions, personalized macronutrient breakdowns, smart food substitutions, and strong security with BCrypt hashing, email‑OTP MFA, and encrypted data protection.
Problem Statement: Traditional logistics systems prioritize cost and time, often neglecting environmental impact. Develop a deep learning model that uses route data, fuel usage, weather, traffic, and cargo weight to optimize delivery routes for minimal carbon emissions, helping companies make green logistics decisions.
Tools used: Streamlit, Pandas, Matplotlib, GeoPandas, Contextily
Objective: An interactive web app that optimizes supply chain routes by allowing users to upload route data, validate it, and visually compare original vs. optimized paths on an interactive map to improve efficiency and reduce environmental impact.
Insights:
The tool visualizes old and optimized delivery routes on a map with clear start/end markers, calculates CO₂ emissions for each path, and highlights total carbon savings. With an easy-to-use interface, it helps logistics teams compare routes, quantify environmental and cost benefits, reduce fuel use, and minimize overall carbon footprint.
Problem Statement: In the agricultural and food processing industries, accurately identifying and classifying different types of rice grains is a critical task. Traditional methods of rice grain classification are manual, time-consuming, errorprone, and often require expert knowledge. This creates challenges in maintaining quality control, automating sorting processes, and ensuring
consistency in rice packaging and distribution.
Tools used: Python, OpenCV, Numpy, scikit-learn, xgboost, streamlit
Objective: To address the challenge of accurately identifying rice grain types, we
propose a lightweight machine learning system that classifies rice
images using Principal Component Analysis (PCA) for dimensionality
reduction and XGBoost for classification.
Insights:
A lightweight ML system that classifies rice grain types by preprocessing images (grayscale, 32×32), reducing features using PCA, and predicting with an XGBoost classifier. Deployed via a simple Streamlit interface, it enables users to upload rice images and receive instant, high‑accuracy grain type identification.
Problem Statement: The AI Helper for Farmers is designed to help farmers detect crop diseases early and
receive basic agricultural solutions using artificial intelligence. Farmers in rural areas often
face challenges such as the inability to identify crop problems promptly, limited access to
expert agronomists, and difficulties using traditional apps due to language or literacy barriers.
This app addresses those issues by allowing farmers to simply upload a picture of the
affected plant and describe the problem using their voice.
Tools used: Python, Pandas, Streamlit, MySQL, ML-CNN, Google API(Speech-to-text)
Objective: This AI-driven project supports farmers in detecting crop diseases. It combines computer vision
and speech-based input for diagnosis. Images are classified using a CNN-based deep learning
model. Symptoms are described via voice. Secure login is enabled. The system dynamically
links diseases to treatment suggestions. Streamlit provides an intuitive, interactive diagnostic
interface. Image preprocessing ensures consistent model predictions. Voice recognition
improves accessibility for non-literate users. Fallback analysis aids diagnosis when no image
is available. Class labels are normalized for clarity and consistency. Model training uses Torch
and handles unseen samples robustly. The tool empowers smart farming with real-time insights.
It’s designed for scalability, accuracy, and farmer-friendly use.
Insights:
An AI-powered web app that helps farmers identify crop diseases through multimodal inputs. Farmers upload leaf images analyzed by a CNN for disease detection, or describe symptoms via voice in local languages, processed using speech recognition and NLP. By combining visual and verbal inputs, the system improves diagnostic accuracy and provides basic, actionable guidance with remedies and preventive steps. Designed for rural, smallholder farmers, it offers an accessible, cost-effective tool for early disease detection and sustainable farming.

