Dr K V N Rajesh, a certified trainer on the Azure platform and holds a doctorate in deep learning, says that his model has achieved 95 per cent accuracy in identifying images of leaves available on the internet, based on the plant “flavia”.
With Ayurveda making rapid strides in the Indian pharmaceutical industry, the winning entry by Dr K V N Rajesh in the first phase of the ongoing Microsoft Azure Blogathon contest may prove to bring dynamic change rooted in this direction. The programme, written in the Python programming language and using Microsoft’s Azure cloud platform, is intended to help distinguish between different types of plants.
Rajesh, who is a certified trainer on the Azure platform and holds a doctorate in deep learning, said that there are many species of plants and identifying them based on the leaves is a task that needs to be solved. “Many companies making Ayurvedic products use leaves in their products, and identifying the exact species is a crucial problem,” he says. He also explains that the same concept can be applied in the agricultural field, specifically with regard to dealing with plant diseases and pests.
“Plant leaves have attributes such as colour, shape, texture etc. based on which you can identify them. I developed an architecture based on a deep learning model, a subset of artificial intelligence which will automatically grasp these attributes. The model which is generated is trained with images of leaves. When the model training is complete, it is validated based on the data set of the leaves,” he points out.
The second phase of testing involves testing the accuracy of the model. According to Rajesh, his model has achieved 95 per cent accuracy in identifying images of leaves available on the internet, based on the plant “flavia”.
“Thirty per cent of the images were used to train the model, and the rest were used to test it,” he says. He intends to take the programme further by incorporating it into a mobile application.
source/content: indianexpress.com (headline edited)