Research

Agricultural Sciences

Title :

e-KRUSHI: PRECISION AGRICULTURE

Area of research :

Agricultural Sciences

Focus area :

Plant Health Monitoring

Principal Investigator :

Dr. Suneeta Veerappa Budihal, KLE Technological University, Karnataka

Timeline Start Year :

2023

Timeline End Year :

2028

Contact info :

Details

Executive Summary :

Towards addressing these issues, the identified gaps are: 1. The timely identification of the nutrient deficiency and diseases detection. 2. The guidance to the farmers to use the required amount of fertilizers and pesticides to meet the expectations of the healthy plants. 3. A tool to identify the nutrient deficiency and diseases accurately without human intervention, i.e., non-invasive mode. Towards addressing these gaps, the identified objectives are: 1. To design and develop a learning framework to detect macro nutrient deficiency and infected diseases. 2. To develop a novel optimized process for improved performance and generate timely response to act on the remedial initiatives. 3. To design and develop a tool for precision agriculture. The methodology is outlined as follows: 1. Literature survey: The state of the art techniques for plant nutrient detection through leaves is carried out in many such as image processing, sensor and IOT based, machine learning and deep learning techniques. In the above methods the ML and DL are the current trends in the precision agriculture. ML algorithms are restricted to supervised and unsupervised learning models and it is associated with limited data sets and accuracy. The deep learning algorithms with the current architectures such as VGG 16, ResNet, GoogleNet, Alexnet the proposed framework is to be experimented. The Google colab, Pytorch and Keras frameworks are also analyzed for further exploration. 2. Data set generation /collection: The Generative Adversarial Networks are developed to generate the dummy image data sets for further training the designed model. The plants are experimented for nutrient and disease detection. 3. Experimentation: The experimentation is carried out using Python as the coding tool and for tool development JAVA scripting is used. The software module is trained and tested for a section of dataset used for training leaf types. The accuracy and error analyses are carried out for all leaves. 4. Validation: The University Of Agricultural Sciences, Karnataka will provide the data using invasive techniques for nutrient detection in their respective laboratories for the same set of leaves and are compared and validated.

Co-PI:

Dr. Saroja V Siddamal, KLE Technological University, Dharwad, Karnataka-580031

Total Budget (INR):

15,68,600

Organizations involved