Sayan Bairagi
Souren Mondal
Krishnendu Ghosh
This project 'AGRIVISION' presents an artificial intelligence-driven precision agriculture platform that synthesizes remote sensing data with inputs from environmental sensors to deliver insights at the field level, enhance decision-making processes, and facilitate proactive management of crops. The proposed framework leverages the Hyperspectral Imaging Library(Sentinel2,copernicus 2), Image Processing Toolbox, and Deep Learning Toolbox of MATLAB to process streams of hyperspectral images using Google Earth Engine. By matching the information of images with historical datasets, the framework infers vegetation- and soil- indices, and it makes it easier to detect the signs of stress and anomalies. Advanced artificial intelligence systems, such as Convolutional Neural Networks (CNNs), as well as Long Short-Term Memory (LSTM) networks, are utilized to process spectral and time-related patterns, thus predicting possible vegetation stress, disease outbreaks, and pest infestations before they become serious. we are building the frontent of our app using flutter. For login purpose to store data we are using Firebase which is online database, and creating a online server using render. In the backend we are modelling our python codes to build our own API and using Fast api for backend of flutter. By coding in Fast api, we created our own API. In that API, we implemented and ran the Python code. This has been deployed on an online server, and through our own API it is running. Using Fast api, it connects with the Flutter backend, and through this we are able to obtain the hyperspectral image, temporal trend plot, and spectral health map. In addition to imaging functions, the platform integrates real-time sensor information of soil and environmental parameters, including soil moisture, air temperature, humidity, and leaf wetness. Integration of the various data sets enhances accuracy, brings context to spectral anomalies, and triggers alerts unique to respective zones for site-specific actions. Integration of spectral imaging and environmental data enables a more reliable identification of stress factors and thresholds of adverse condition leading to crop degradation or pest infestations. To assist usability and accessibility, we created the Dashboard based mobile application to serve as the platform's user interface. Its dashboard also houses essential functions, including: Meteorological forecasts incorporating local temperature and humidity changes using open-weather API. Predictions of Crop Health with AI-driven suggestions Spectral Health Maps of areas exhibiting stress-free and risk-prone regions Field-level soil condition soil condition reports Pest Risk Assessments for on-schedule preventive measures Visualization of hyperspectral data for detailed spectral analysis. Report Cards and Summary for record-keeping and analysis Instant notification and alert for serious issues. The major users of this platform include farmers, agronomists, researchers, and field technicians who require real-time, site-specific information to make management decisions. With the delivery of mobile-compatible reports, interactive dashboards, and actionable alerts, the platform helps users make decisions from data that improve crop health and foster sustainability. This project ultimately converts crop surveillance from a reactive process, wherein problems are solved after visible harm, to a predictive and preventative system wherein risks are anticipated and averted before they become too late. Through a combination of AI, hyperspectral imaging, and IoT-enabled sensing, the platform helps usher in precision farming and enables responsible farming. Overall, the developed AI-based farm monitoring platform is a scalable and adaptable solution integrating emerging imaging technologies, machine learning techniques, and sensing technologies. It delivers time-responsive, accurate, and actionable information at the field-level to reduce risks, enhance yields, and provide forsmarter farming practices in the future.
Agriculture is subjected to continuous stress from land degradation, uncertain weather, and pest infestation. All this results in decreased yields and economic losses for farmers. The conventional monitoring systems are usually slow, inconvenient, and less accurate. To address this, we envision a consolidated software platform that combines remote sensing (hyperspectral imaging) with field sensor data to provide timely information about crop health, soil status, and pest threats. By processing imaging data with environmental parameters, the system allows early stress factor detection and enables farmers to transition from reactive to proactive crop management. our platform is based on MATLAB's hyperspectral imaging library, Image Processing Toolbox, and Deep Learning Toolbox. It analyzes hyperspectral image sequence, registers them with historical datasets, derives vegetation and soil indices, and employs AI models like CNN and LSTM to identify trends and forecast vegetation stress or disease susceptibility. Sensor readings such as soil moisture, temperature, humidity, and leaf wetness are combined with spectral characteristics to improve accuracy, identify anomalies, and initiate zone-based alerts through the software. In order to make these insights convenient, we created the Agri AI Dashboard mobile application 'AGRIVISION'. The application gives users(Farmers, Students, agronomists, researchers): Weather Forecasts (temperature, humidity) Crop Health Predictions with suggestions temporal trend plots Spectral Health Maps indicating stress-free or risk areas Soil Condition Reports Pest Risk Alerts Hyperspectral Data Visualization Downloadable Report Cards Instant Alerts for major issues This dashboard allows farmers, agronomists, researchers, and students to interact with the platform via an easy-to-use interface, view in-depth analytics, and access mobile-responsive notifications and alerts for fast response. So, this solution brings together AI-based image analysis, environmental sensing, and a simple app interface to facilitate precision agriculture. It facilitates sustainable farming by allowing early detection, localized decision, and proactive crop management.