Computer Vision Platform for Plant Species Identification and Precision Agriculture

Plant species and sub-species identification with the help of computer vision

Objective

• A Computer Vision model will be developed to a) Identify the species as well as sub species of the cultivated crop, b) detect presence of weed density c) ascertain plant age and stress.

Description

• As the crop yield and quality are impacted by soil health, irrigation levels, application of pesticides and herbicides, weed growth and pest infestation, computer vision models based on the images of the plants and information on aforementioned parameters will be collected and measured to respond to the following questions: Can we identify the crop and its subspecies?, From the plant density can we detect the presence of weeds?, Is the crop healthy?

Impact

• The predictive models will enable the farmers to determine the necessary interventions to optimize or maximize crop outcome under the given conditions. On further enhancement with deep learning methodologies, the developed application will also determine the correlations between the growth of crops (paddy in the first instance) and climatic conditions, nutrition availability at various times as well as presence of living organisms in the vicinity of the crops. It will be suitable for both the modern system of agriculture using chemical inputs as well as the traditional system using organic or natural inputs. It is envisaged that over time the knowledge gleaned from the system will help the farmers transition to precision agriculture using predominantly organic inputs.