Discover, Connect & Collaborate at TECHINNOVATION 2021
Video Analytics for Plant Growth & Disease Detection
Crop production is one of the major sources of income in the world and more than half of our population depends on agriculture for livelihood. Crop cultivation is a tedious and laborious process. Diseased crops are devastating to the farmer. One of the major concerns is the lack of knowledge and manpower to identify diseased crops at an early stage to prevent the spread of infections to healthy crops.
This technology offer is a fast and accurate, artificial intelligence (AI)-based computer vision system that can detect plant diseases. This system is more reliable and scalable than the traditional, manual methods in detecting plant diseases. It uses video analytics, which has algorithms that automatically analyse large collections of images and identify features that can be used to categorise images with minimum error. Images of plant leaves are digitally pre-processed to get clear, noiseless enhanced leaf images. These enhanced images are then used for analysis to detect diseases. Generally, plant leaf image colour and texture are unique features, which can be used to detect and analyse the diseases.
Technology Features, Specifications and Advantages
The traditional method of checking for diseases in plants is through visual checks, but this method is not consistent, and is inefficient in detecting the diseases associated with plants. A faster and scalable alternative method is the use of computer vision which can be more reliable and productive. This technology uses computer vision and video analytics to evaluate the probability of each disease for plants of any color. There are many variations of the same disease among each class of crops; every disease presents specific characteristics that make them different from others, e.g., colour, texture, shape. These characteristics and their various combinations are used by the algorithm to determine the specific disease present.
The trained model detects diseased plants within acceptable confidence. The research and development efforts have the potential to be translated into IP in the following areas:
- Once co-related with real-time data, it can be a predictive tool to enable farmers to “predict” the probability of the plant disease with a known accuracy.
- Farmers will also be able to formulate suitable interventions through medications or other means. A detailed study with control groups will be useful to determine the effectiveness of said interventions.
- Further developments with appropriate refinements to the type, suitability, sensitivities, precision and accuracy of the detector may enable early detection of other diseases.
- The device will also be useful as an early warning detector for plant disease infections.
- Early intervention to prevent the spread of plant disease, and hence improve overall yield.
- Reduction of manpower and labour required to do visual checks.
Currently, the proof-of-concept stage is completed, and this technology is available for project collaborators or consultancy projects to field-test the technology.