Random Nets for Dalbergia Sissoo Disease Identification

Dalbergia sissoo (locally known as ?Sheesham? in Pakistan and ?Tahli? in Punjab) commonly known as North Indian rosewood. It is a medium to fast growing tree having highly impact on economic as well as local due to using in Timber, Fuel Wood and Teeth Brushing etc. As time passes number of various d

Project Title

Random Nets for Dalbergia Sissoo Disease Identification

Project Area of Specialization

Artificial Intelligence

Project Summary

Dalbergia sissoo (locally known as ‘Sheesham’ in Pakistan and ‘Tahli’ in Punjab) commonly known as North Indian rosewood. It is a medium to fast growing tree having highly impact on economic as well as local due to using in Timber, Fuel Wood and Teeth Brushing etc. As time passes number of various disease like die-back, wilt etc. attack on Sheesham and there is no highly convenient mechanism to identify the root disease. Early approaches used to identify disease from naked eye which is traditional method. A recent approach used digital image processing for the disease identification and classification of affected leaves of Sheesham. Deep learning approach used for the disease identification giving better accuracy from all previous approaches. Convolutional Neural Network detects and diagnoses Sheesham disease with best performance when trained. However on larger dataset time and computational power is a bottle neck problem. It is necessary for CNN to train for disease identification but time cost factor never negligible. All previous approaches used for Sheesham disease and other plant leaves identification giving better accuracy from one to another but no one focus on time complexity. Deep leaning producing effective result with lower time complexity from all other approaches. But on lager dataset it consumes longer time. Our approach Morphnet that is scalable and fast by reducing time. It used pruning algorithms maintain the core theme. As time and computational power both core part of project. It reduces both the computational power as well as time. FLOPS Floating Point Operation per Second are used to compute the computational task. We are interested to design a Neural Network that minimizes the cost function and prone optimal solution into number of hitting the target data. A Morphnet regularizer is used to providing threshold value in term of when there is no layer. It is scalable over larger dataset and learns from the growing data to improve time complexity. Morphnet reducing the time and resource without scarifying accuracy.

Project Objectives

Reduce cost and Energy

Project Implementation Method

Tool Colab

Benefits of the Project

?Having large no of processing layers, work on Neuron

?Step ahead of AI

?CNN work on complex process and perform pattern recognition on large data set

?Improved version between real time and already captured image

Reduce Time and Energy

Flexible and user friendly

A complete Agronomist without doing Degree

Technical Details of Final Deliverable

Used Colab and colletion of image data set of Dalbergia sissoo to identify the specific disease via image segmentation.

Final Deliverable of the Project

HW/SW integrated system

Type of Industry

IT , Agriculture

Technologies

Artificial Intelligence(AI)

Sustainable Development Goals

Required Resources

If you need this project, please contact me on contact@adikhanofficial.com