Thapar Institute of Engineering & Technology on 19 January 2018, signed a comprehensive institutional agreement with Israel’s Tel Aviv University (TAU), covering research, teaching, and student exchange between both institutions.
Under this initiative TIET-TAU Center of Excellence for Food Security (T²CEFS) has been established. The various cutting edge research projects are undertaken under this Center of Excellence including Digital Village: A Data-Driven Approach to Precision Agriculture in Small Farms. This project is running under the leadership of Prof. Yosi Shacham, Professor Electrical Engineering, Faculty of Engineering, Tel Aviv University, Israel and Director TAU-TIET Canter of Excellence for Food Security and Dr. Moushumi Ghosh, Professor, Department of Biotechnology and Coordinator TIET-TAU Center of Excellence for Food Security.
Artificial Intelligence
Based Digital Village System
Prime Objectives
- To establish the First pilot project in north India (Punjab)as an efficient and sustainable agriculture model with the help of modern engineering AI tools.
- To collect season-wise soil and crop information through a manual survey of respective farmers of the field starting from the time of sowing to harvesting such as the amount of manure, pesticide, herbicide, water, yield, etc.
- To Design and Develop a survey mobile app for collecting field data from farmers by conducting weekly, monthly, and yield surveys.
- To Capture the visual data of crops for weed detection and crop disease identification.
- To Build an AI-based system for automatic detection of weed and crop disease detection systems from the visual data captured from the fields.
- To Sense environmental parameters like soil moisture, temperature, pressure, and luminosity of a crop field by using IoT sensors.
- To Develop deep learning models for forecasting environmental parameters.
- To Capture and Analyze remote sensing data by using open source APIs for precision agriculture.
A Survey Mobile App
- Collected onsite data from the field of the farmers.
- A survey conducted by a field engineer.
- A Graphical user interface (GUI) based (android) application developed to conduct surveys in the regional language of the farmers under study.
- Several surveys conducted in farm fields to capture before plantation and after plantation details of the field. These surveys were conducted in a weekly, monthly, and seasonally fashion (crop-seasons) on a time-stamp basis.
- The field engineer also helped to collect manual visual data related to crop by using mobile cameras.
Visual data of Crops for Weed Detection and Crop Diseases Identification
- The proposed system captures the visualized images of the crops and the fields along with its demographic parameters such as latitude and longitudes of the location.
- The proposed system also captures the data at regular intervals using installed HD cameras.
- These values are mapped with the impacted site of the field.
- We also explored the use of automated carts for the collection of weed visual data, as these carts will run on the ground and will help to capture pictures of small weeds.
Detection of Weed and Crop Disease Detection Systems
- An AI-based system developed after the acquisition of the visual data of the crops that caters the weeds detection and analyzed the diseases of crops of wheat and rice.
- The AI model required samples of the observations for the training. These samples consist of visuals of the aerial surveys as well as captured images through mobile phones.
- The captured data may have more than a category of weeds and diseases.
- A deep learning-based algorithm will be used for the detection of weeds and diseases from these image-based datasets.
- The model classifies the images of the plant whether the plant is affected by the weeds or diseases or not.
- A comparison was also computed to analyze the accuracy of the trained model by using specific performance metrics like TP, FP, TN, FN, Precision with, Recall, and F-Measure.
- The experimental data will be evaluated on the AWS, Floyd hub servers.
- The obtained results were also compared with the real-time captured data of the crop and the soil to estimate the precise detection of the weeds and the diseases.
IoT sensors
- The Libelium hardware was used to acquire information about the soil properties, i.e., soil moisture, soil temperature, pressure, luminosity, etc.
- This would lead to the preparation of an informational chart about various soil properties and behavior of crops.
- Various agricultural sensors such as soil moisture, humidity, and temperature would be installed on the testbed.
- These sensors collect the readings for an entire agricultural season.
- Furthermore, the data was uploaded on a cloud server for enhancing the flexibility of data, disaster recovery in case of system failure, and to cater to the needs of accessibility of data from any location.
Forecasting Environmental Parameters
- The online prediction of the estimated soil moisture, temperature, humidity, and luminosity.
- Date-time-based forecasting of these readings were evaluated using the time series analysis.
- ARIMA, LSTM, and the PROPET were the prominent models to perform the time series analysis.
- The regressive dataset was processed using these models and the best or the hybrid model was proposed to acquire the best results of the regressor model.
- The results of these models were evaluated using different performance metrics such as MAPE, MAE, MSE, and R2.
- This analysis will help the farmer to better plan the irrigation for better utilization of natural resources.
Capture and Analyze Remote Sensing data
- Google Earth Engine is a platform for scientific analysis and visualization of geospatial datasets, for academic, non-profit, business, and government users.
- Earth Engine hosts satellite imagery and stores it in a public data archive that includes historical earth images going back more than forty years.
- Earth Engine also provides APIs and other tools to enable the analysis of large datasets.
- We also analyzed the water coverage, land-use change, vegetation indices, and assess the health of agricultural fields, among many other possible analyses.
- Apart from this, these fetch observations were co-related with the acquired data from the IoT-oriented device.
- Thus, we also compared the grounded data with the satellite data which will be beneficial to analyze the instant relation between the soil and the crop.
A Real-Time Web-System
- All the information analyzed through these AI-based models are used to develop monitoring tools both web-based and mobile-based that can be used by the farmers.
- All these interfaces will have the support of the local languages so that farmers can use this with ease.
TIET-TAU Center of Excellence for Food Security T²CEFS
Organization
TIET-TAU Center of Excellence for Food Security T²CEFS
Address
P.O. Box 32, Bhadson Road, Patiala, Punjab, Pin -147004, India